Image-Guided Aspirations and Biopsies

Brief Historical Summary of Image-Guided Biopsies.

When biopsies were first performed in the 1930s through the 1960s, they were performed at the bedside by clinicians without imaging. At that time, the selected organ, usually liver or lung, was located by physical examination. Entrance sites were chosen using palpation and auscultation with a stethoscope. The skin was prepared with antiseptic, but local anesthesia was not always used. The biopsy was performed quickly during a single breath hold, the theory being that if the needle was inserted and withdrawn quickly enough the patient would not feel it or would not move. Laboratory measurements of hematocrit and coagulation studies were not used frequently. Complications were frequent.

It is almost amusing in retrospect to reflect on two of the most important historical advances for unguided liver biopsies. First, it was discovered that the lateral intercostal approach was found safer than the anterior subcostal approach. With the clear knowledge of anatomy today, one reflects why they did not realize that the other contiguous organs, such as the stomach, gallbladder, colon, duodenum, and pancreas, could potentially be damaged by this path. Imaging guidance precludes this possibility. Second, the liver biopsy was proved safer when performed while the patient suspended respiration and motion. One eminent physician of the time, Menghini, gained great notoriety because he had the patient suspend respiration, and the needle was thrust in and out during 1 second. This practice prevented motion of the diaphragm and reduced the number of complications. His other great innovation was having a single physician apply suction with a syringe while inserting the needle; previously, one person applied suction by means of an attached tubing while the other inserted the needle. It is very strange to note that some physicians today are proposing the “old Menghini” two-person technique of the 1940s where two people perform a biopsy. One person does the needle work and the second provides vacuum with an attached plastic tube. (To each his own!)

Relative to guided procedures, Lundquist was the first to show the usefulness of “guided” biopsies using US and FNA; the sample was cytologic. The US for guidance was the B mode (before real time), which was used to mark the skin surface; no direct visualization of the needle was possible.

Lundquist's reported experience showed conclusively that this method was superior to “blind” procedures. He also reported the use of the 18-gauge Menghini needle (end-hole cutting needle compared with the side cutting needles of today) for selection of the needle entrance site. In a series of 1000 biopsies with the Menghini needle and 2611 procedures with the “skinny” needle, there was a very low complication rate; one hematoma occurred with the skinny needle. For perspective, one should realize that in the 1970s, the imaging technology was such that only extremely large and numerous lesions could be visualized and even considered for biopsy.

FNA Versus Core Cutting Needles.

From today's perspective, it seems inconceivable that any controversy about the effectiveness or specific benefit of FNA versus core cutting needles existed, but at the major meetings of the 1980s and 1990s, debates about skinny needles versus large cutting needles were extremely controversial and charged with emotion. Proponents of skinny needles used references from the unguided biopsies of the antiquated days of the 1950s and 1960s to counter the benefits of larger needle samples.

With the unguided biopsies, a variety of core biopsy needles were designed and evaluated; a brief summary is important because invariably vendors reintroduce different permutations of these needles periodically using a different name. A prospective series in 1978 by Perrault and associates showed no complication difference among the Vim-Silverman, Menghini, and Tru-Cut needles. The comparison by Bateson and colleagues in 1980 of the Tru-Cut needle and the Menghini needle (end cutting needle) showed no diagnostic or other complications, although the physical integrity of the sample was more consistently intact with the Tru-Cut needle, especially with cirrhotic livers. Large cutting needles were not used routinely in imaging until the 1980s, when our group published numerous articles about the basic characteristics of needles, the diagnostic yield of larger needles, and the comparison of outcomes with skinny and cutting needles. In the 1980s, we were accused of “malpractice” for using such needles because of the “reputed” dangerous complications they posed for patients, and yet today we believe some physicians may overuse cutting needles and do not follow the guidelines similar to ours, which ensures a low complication rate. It is ironic, looking back at the early days of interventional procedures, that it took 2 intense years of reviews and negotiations to get our first paper on lung biopsies and cutting needles by Goralnik and coworkers published, and now such procedures are commonplace and perhaps overused (see “ Lungs and Mediastinum ” under the heading “ Selection of Cutting Needle Type and Caliber ,” later). We initially proposed the routine use of such needles in 1979.

Since the 1990s, the superiority of cutting needles has been repeatedly documented and confirmed. Such large needles are used preferentially in all organ systems, with excellent results and acceptable complication rates. Having championed their usage, we remain enthusiastic, but we believe that careful choice of patients, lesions, anatomic approaches, and use of contrast material is important to balance the benefit and risk for each patient. With proper techniques, the diagnostic yield approaches 95% and the complication rate is less than 1%. Some physicians use cutting needles without appropriate patient selection and precautions (see later sections).

Advantages.

CT uses a beam of radiation to penetrate virtually all natural and artificial materials within the human body. Information concerning attenuation of these substances can be measured and an image can be reconstructed. Because of the wide latitude of imaging capabilities, an extremely low-density material (e.g., gas) can be imaged, as can high-density materials such as metal, bone, or synthetic catheters ( Fig. 67-5 ). In addition, CT is sensitive enough to detect subtle abnormalities such as fluid densities, cysts, or gas-containing small pockets or fistulas ( Fig. 67-6 ).

The ability to image high-density material accurately offers wide versatility so that any type of instrument (i.e., needle, drainage tube, or other device) may be used in a procedure. Contrast material can be injected when vascularity, an anatomic space, or an abnormal cavity needs further delineation. An abscess can have unpredictable serpiginous tracts or communicating channels that can be difficult to perceive unless contrast material is used ( Fig. 67-7 ).

Because the CT unit uses an air interface, it consistently provides detailed and complete image even in patients with skin issues. In patients who have draining sinuses, wounds, ileostomies, scars, or other unusual surface problems, the examination can be performed without difficulty. Another advantage is that the longitudinal scan (e.g., Topogram, Scoutview) that we originated is now available on all CT scanners, making it possible to obtain conventional radiographs of the anatomy and to visualize instruments or catheters during insertion or manipulation.

Finally, real-time CT fluoroscopy can be used for cases of very small mobile masses or abnormalities. With the evolution of patient referral, lesions sent for biopsy have become smaller and smaller, testing the skills of even the most experienced radiologist.

Accuracy of localization.

The potential accuracy of the CT-guided method can best be appreciated by examining a schematic drawing of a cross-sectional image (see Fig. 67-1 ). If the instrument is placed within an anatomic area and the CT scan shows the needle in that area, the needle must lie within the 10-mm section of the abdomen corresponding to the CT section. Location of an instrument in this plane is limited by the section thickness, which is 10 mm; the resolution in the z axis is +5 mm. If one uses narrower slices, such as 5 mm, the accuracy in the y-z plane is +2.5 mm. Resolution of the CT system in the x-y plane is less than 1 mm with modern scanners. The precision required to determine the position of such an instrument is less limited by the resolution of imaging equipment than it is by the degree of the operator's accuracy, as determined by manual dexterity and by the degree of accuracy of the guidance system, rather than by the imaging equipment. Because of the increased detection of smaller lesions, incidental findings must be managed, and a definitive biopsy would be the best alternative; robotic assistance that positions the device during real-time fluoroscopy would be required.

The ability to localize the needle tip consistently and accurately in an axial plane is a distinct advantage of CT over fluoroscopy or US. Although fluoroscopy can provide multiplanar visualization of a needle or instrument, it is incapable of directly visualizing soft tissue abnormalities except in the chest. Fluoroscopic localization of instruments depends on indirect information obtained from displacement of contrast-filled structures (e.g., bowel, biliary).

The special advantage of CT guidance is that visualization of the needle is a foregone conclusion, and special technical approaches or concerted efforts are not required to visualize the needle. CT is not as sensitive for subtle lesions, however, and occasionally the combination of MRI and CT guidance can prove worthwhile ( Fig. 67-8 ).

Versatility.

Versatility is still one of the most valuable advantages of CT-guided procedures. There is significant versatility in selecting the entry point for the biopsies because of the completeness of the CT image: a full 360-degree view of the patient is provided. All organs, including the bowel, are visualized so that a variety of needle pathways can be chosen. It is possible to choose appropriate pathways to avoid specific organs or the bowel if clinically indicated. Changing the body position of the patient may be advantageous for opening windows for instruments because of organ shifting or patient comfort (changing positions creates no adverse effect on image quality). Finally, a wide variety of instruments can be used, ranging in size from a 23-gauge Chiba needle to a 14-gauge Tru-Cut needle. Of all of the advantages of CT procedures, versatility is the one that is least recognized.

Varying the patient position is useful for patient comfort or for planning and execution of the procedure ( Figs. 67-9 to 67-11 ). As noted in the subsequent clinical sections, three major benefits are derived. First, if patients are going to cooperate during a procedure, it is crucial that they are comfortable on the table. Patients with back pain or localized discomfort in the flank can be induced to remain still during a procedure if they can become comfortable by shifting their position. Permitting the patient to lie in a prone, lateral, or oblique position can go a long way toward improving patient comfort and cooperation and success of the procedure.

The second benefit of moving patients into different positions is that the internal or external anatomy shifts accordingly and permits better and safer trajectory planning and execution. For chest procedures, such movement may permit displacement of the scapula, making a posterolateral approach easier (see Fig. 67-9 ). Within the chest, lesions in the anterior mediastinum can be sampled more safely because positioning patients on their side may shift the anterior junction line sufficiently to permit a pathway avoiding the lung parenchyma (see Fig. 67-10 ). In the abdomen, bowel and organs shift sufficiently to permit a safer sampling procedure.

Third, in some cases when a patient has difficulty taking reproducible breaths and the lesion is mobile, placing the patient in a lateral or oblique position can help in limiting the movement of the lesion. When the ipsilateral side of the patient is dependent, the weight of the viscera partially immobilizes the diaphragm and limits movement of the lesion (see Fig. 67-11 ) even when it is in various precarious sites, such as adjacent to the spleen.

It is perplexing that some radiologists do not change the patient position or trajectory of the needle to avoid uninvolved structures. By varying the patient position slightly, it may be possible to produce movement of a loop of bowel, permitting an anterior biopsy without traversing bowel ( Fig. 67-12 ). In other cases, one can approach retroperitoneal lesions posteriorly through the flank instead of traversing bowel or other structures by using an anterior approach. Although traversing bowel is seldom a problem, in selected cases, secondary infection may develop from a microcontamination ( Fig. 67-13 ).

Occasionally it is beneficial with simple and complex procedures to perform 3D imaging with axial, coronal, or variable planes. For unusual biopsy procedures or ablations, it may be beneficial to determine the exact location of a cutting needle before acquiring the tissue sample ( Fig. 67-14 ). In complex drainages, such as rectal drainages ( Fig. 67-15 ), 3D imaging is well suited because it is difficult to ascertain the precise trajectory of the needle when the approach is made through the rectum. Also, the precise placement of the catheter in the abscess cavity can be confirmed.

For the new ablation treatments, 3D rendering is especially important because accurate planning and execution of a treatment plan is crucial for consistently successful procedures. Accurate placement is necessary to ensure that all portions of the tumor are treated and there is adequate overlap of the treatment areas. It is of further benefit to confirm that no adjacent uninvolved anatomy is being damaged ( Fig. 67-16 ).

Disadvantages.

CT guidance has three disadvantages that must be considered when planning image-guided procedures: (1) use of radiation, (2) less sensitivity for imaging primary liver tumors, and (3) less sensitivity for imaging musculoskeletal tumors.

The CT scan uses radiation for generation of the scans, but the dose delivered to the patient is much less than with other specialized studies, such as angiography. With modern MDCT scanning devices, the detector systems are remarkably efficient, using almost 100% of the dose. Although excessive doses should be avoided in all circumstances, a single scan has a sufficiently low level that in cases of acute illness that may jeopardize a patient (e.g., appendicitis) a CT scan and procedure may be justified. In the more recent generation of scanning devices, as a result of increased awareness about radiation safety, vendors are producing “photo-timed” CT scanners. The detector systems are being used in conjunction with the computers to measure the dosage “on the fly” so that optimization of dosage occurs even within the context of a specific scan. I cannot avoid the opportunity to say, “I told you so and it's about time,” noting that this was proposed many years ago in our second edition of the textbook, after our first article on CT dose reduction in 1981 and subsequent recommendations to use the lowest possible dosage for image-guided procedures.

There is a definite lower detectability rate of tumors in the liver and muscle tissues. Because primary tumors of the liver are sometimes difficult to see, one should perform procedures in conjunction with real-time US or reference an available MRI scan (see Fig. 67-8 ).

Advantages

Instrument localization.

US has gone through several periods of great enthusiasm as technology has improved. In the early years of interventions, the metal instruments were difficult to visualize because the needle completely reflected the US beam, preventing sufficient return of echoes to permit visualization. The sensitivity of new US equipment is so great that it is easy to visualize the tip and pathway of needles in the range of 20 to 18 gauge ( Figs. 67-17 and 67-18 ). The very slight motion of the needle during insertion is well visualized, permitting the “trail” of the needle to be seen clearly. Alterations in the “surface” characteristics of the needle sides and the tip show the end of the needle well. Any tiny gas bubble that is adherent to the end is displayed well.

Disadvantages.

There still remains significant difficulty in special circumstances when bone or gas is present either before or during a procedure. Although US may be used for single or small instrument placement for radiofrequency ablation (RFA) or cryoablation, it is limited when numerous probes must be inserted and localized. Assessing treatment in such cases is difficult because the iceball or gas generated may not be completely assessed or imaged.

Advantages.

MRI advantages relate to its great sensitivity for displaying subtle changes in tissue characteristics, permitting instrumentation. This sensitivity enables visualization of subtle abnormalities not seen as clearly on CT or US (see Fig. 67-8 ); this is frequent with primary liver mass lesions and some cystic tumors of the pancreas.

MRI can survey such nodules and determine which are most suspicious for change to neoplasm. Differentiation of dysplastic nodules compared with hepatocellular carcinomas usually can be made by quantitating T2 differences (see Chapter 43 ). Also, when ablative procedures are performed, the “real-time” damage or change to tumor tissue can be determined by change in the signal. Another example of superior diagnostic sensitivity is in pancreatic tumors (see Chapter 46 ). MRI is remarkably superior to CT for the detection of subtle islet cell tumors and the details of some cystic tumors of the pancreas.

Because of these qualities, MRI is well suited in selected cases to follow the progress of the treatment procedures. It has been well documented that MRI is the most accurate modality for displaying the tissue changes from RF treatment, as noted by Lewin and coworkers and Merkle and colleagues in numerous reports ( Fig. 67-19 ). Experimentally, MRI has been able to display characteristics correlating with absolute tissue temperature. With the current emphasis on treatment, it also may be possible with higher-field magnets having broadband capabilities to visualize drugs in situ as they circulate within the tissues and the target molecules. It is not illogical to assume that in the future, these qualities will be exploited by interventional strategies, most likely therapeutic approaches, if issues related to accurate and convenient instrument placement can be resolved.

Another advantage of MRI is its ability to image blood vessels and to assess vascularity as it relates to general blood flow. MRI during standard pulsing sequences shows vessels even without contrast material. Moderate-sized vessels are clearly seen as a flow void. Although the anatomic spatial resolution of MRI is less than that of CT, it can also display specific blood flow characteristics, as one might see in hemangiomas, using gadolinium-enhanced dynamic contrast enhancement (DCE). An indirect advantage is that gadolinium has a much lower idiosyncratic rate than iodinated contrast material, so that patients who are allergic to traditional contrast material used with CT can be imaged and evaluated by MRI flow studies.

Because of the proven efficacy of MRI-guided procedures, we have included a complete chapter on the topic in this sixth edition. Chapter 66 —“MRI-Guided Interventions” by Dr. Sherif Nour—provides an excellent comprehensive review, and the subject will not be discussed further in this chapter.

Skinny Needle Complications.

There is no question that the skinny needle is the safest instrument available, but even this needle can produce complications when proper patient selection and techniques are not used. In a careful review of reports of skinny needle biopsies of the pancreas, one can find instances of pancreatitis, pancreatic fistula, cholangitis, needle-tract seeding, significant hemorrhage, abscess formation, and fatal pancreatitis. Such complications were usually the result of incorrect patient selection or choice of anatomic pathway and an excessive number of needle passes. Each organ system has specific issues, as noted and explored subsequently under each of the organ system sections below, but general factors can be mentioned.

Improper patient selection and evaluation of anatomy and vascular properties have produced numerous hemorrhagic complications and deaths. Reported fatalities from liver and adrenal hemorrhage occurred with vascular hemangiomas, hemangiosarcoma, severe coagulopathy, and pheochromocytoma. One can see in reviewing the literature that most complications have occurred in cases with four or five needle passes.

The impact of numerous needle passes with skinny needles on complications was reported by Smith of the University of Massachusetts. Smith reported a survey sent to 100 university hospitals and every fifth hospital in the United States, which solicited information about complications and risks from FNA biopsies. In this report, he noted the occurrence of 4 deaths, 27 cases of hemorrhage, 3 occurrences of needle-tract seeding, and 16 cases of generalized infections (nonfatal). Smith stated, “It is logical to assume that the number of complications should be related to the number of needle passes made during the biopsy procedure. However, serious and even fatal complications, although rare, can and do occur and it is important to be aware of the possibility and take all the appropriate precautions in order to reduce their incidence.”

Even with extensive experience and caution, hemorrhage may occur with FNA aspirations ( Fig. 67-27 ). In most instances, such events are rare if proper methods are used.

Cutting Needle Complications.

One has a greater probability of incurring possible complications with large needles because a large “hole” is being produced. With proper patient selection, proper target and trajectory choice, and meticulous technique, however, the potential risks of these needles can be offset, and very low complication rates can be achieved. Considering the previous controversies in the field about using large needles, we have some concern that such needles are used too frequently by some groups. One can use very large needles safely with good results, but one must pay careful attention to meticulous technique, or serious and fatal complications can result.

The new automated cutting devices are no safer than the standard “manual” Tru-Cut needle that has been used for many years. Some authors have asserted that the rapid speed of the cutting action may lessen local damage, but this concept is nonsensical. In the first report on the biopsy gun, the authors described a serious hemorrhagic complication after a liver biopsy. Potential problems are minimized by CT guidance and a careful, meticulous technique such as I have described. These measures can eliminate many potential causes of complications (e.g., hemorrhage or inadvertent penetration of an uninvolved organ) and can result in excellent accuracy with few problems.

If one uses a large-core biopsy needle with an automated mechanism, it is best to avoid the single-action devices and use instead double-action devices. Double-action devices permit confirmation of the cutting gap in the anatomy before tissue cutting occurs. Critical biopsies can be performed safely because the device cuts only where the “gap” is seen on the scan. If a patient wiggles or moves slightly, the location of the needle and the cutting action are not altered. Single-action devices “cut” the tissue with a single action by the operator; if a patient moves, the anatomy may move a vital structure into the cutting site, and catastrophe may result (see “ Lung and Mediastinum ” for an example of a patient in another institution who died because the patient moved slightly in such a case).

Abnormal INR.

For an abnormal INR, there are two options depending on the time issues involved. If the patient is taking a warfarin-like drug, it should be stopped. The patient can be given vitamin K, which enhances recovery of the prothrombin level to normal over several days, but the laboratory values must be rechecked until at a normal range. If correction of vitamin K is ineffective, fresh frozen plasma (FFP) may be administered intravenously before the procedure. Repeat INR values should be drawn until a normal level is reached. Some authors recommend administration of several units before performance of the procedure, without repeat measurement of the laboratory values, but we believe a repeat coagulation screen is important to ensure that factors have returned to normal.

For patients whose coagulopathy cannot be corrected by the aforementioned means, or in emergency circumstances that require early performance of the procedure, we have developed a new method to prevent hemorrhage, which involves the local injection of blood products as described later under “ Local Infusion of Blood Elements for Hemostasis .” This is not an FDA-approved technique, but it is legal under current statutes; we have used it for several years with excellent results and no bleeding complications to date in more than 50 cases.

Low Count or Impaired Platelets.

With platelet abnormalities—either absolute count or dysfunction—several corrective approaches are possible. If the platelet impairment is simply due to drug administration, cessation of the drug and waiting an appropriate period is appropriate. If drug cessation does not work, a transfusion of platelets can be given. If the absolute count is depleted or an untreatable dysfunction exists, IV platelets should be given. For an impaired platelet function or low platelet count, three possible approaches can be taken: IV systemic platelet replacement, infusion of desmopressin, or local platelet injection in the pathway.

For impaired platelet function, especially for elevated blood urea nitrogen (BUN), desmopressin may be administered. The dosage and regimen for infusions should be obtained from the local pharmacy. Although the exact mechanism of action is unknown, it has been shown over many years that administration of this medication improves the adhesiveness of platelets and minimizes the probability of hemorrhage. The medication is typically most effective on the first administration and is less effective when later doses are administered.

In many patients, platelet impairment is produced by medications such as aspirin, other analgesics, or antiplatelet agents given for cardiovascular reasons. Administration of platelet inhibitors or aspirin has become established as the mainstay of cardiovascular treatment. The significance of this platelet impairment was well described in a laboratory model by Gazelle and colleagues, who studied the effects of aspirin. Patients with impaired platelet function on the basis of either medication such as aspirin or other inhibitors are best treated by stopping the medication and waiting the appropriate time for the effect to resolve. Although the effects of aspirin are long-lasting, we consider a 7-day period without medication as the optimal time interval before a procedure should be performed. Nonaspirin pain relievers are usually shorter acting; a waiting period of 1 day after cessation is usually adequate, but the pharmacy should be consulted for specifics of any medication.

Traditional Infusion Replacement.

The traditional approach for correction of coagulopathy is to administer IV platelets or FFP over the course of many hours to obtain a systemic improvement of these factors. If replacement platelets are used, the infusion should be continued until the platelet count reaches greater than 50,000/µL. Restoration of the INR to acceptable levels is accomplished by IV administration of FFP. Numerous units are given until the INR reaches 1.3, as noted earlier. Fresh plasma permits this restoration of the INR because it replenishes numerous factors such as prothrombin and factors VII, IX, and X, which are produced by the liver and are deficient when liver function is impaired.

There are several difficulties with the traditional method. The most difficult issues are related to obtaining the blood products, administering the products, rechecking the tests, and transporting the patient to the imaging site to perform the procedure. Regardless of the institution, anywhere in the world, there are always logistical problems.

If patients have cardiovascular compromise, the multiple units infused may cause cardiac overload and decompensation. In some patients, there may be antibodies or unknown reasons why such products do not correct deficiencies. If patients show a lack of response to such methods or the clinical scenario is so dire that the procedure must be performed emergently, one can opt for the new method of local infusion of platelets, as described next.

Local Infusion of Blood Elements for Hemostasis.

We continue to refine and study our technique for treating bleeding, the local infusion of blood products. This is not an FDA-approved use of blood products, but under current legal guidelines, a physician can act on the best behalf of a patient in unusual circumstances. If a product is approved for human usage, a licensed physician can use that product for other uses when the benefit to the patient seems to outweigh the risks. Under these guidelines, a physician can use the technique described later in this section, but he or she takes on any associated ethical or legal risks.

The innovation of this technique evolved from the treatment of a radiologist colleague who had refractory leukemia unresponsive to systemic platelet infusion. Having received many platelet infusions over the course of his treatment, he developed antibodies to platelets and hypersplenism. This condition was so severe that if he was given 10 U of platelets, his body would consume the platelets and his platelet count would never increase above “0-10.” After developing a severe spontaneous hemothorax causing compression of his heart and diaphragm, he was refused treatment by thoracic surgery and was sent to us for percutaneous treatment. We recognized the possibility of rebleeding but consented to perform the procedure for temporary relief.

At the time, we reasoned that one might be able to use platelets with local administration as an alternative method. The rationale was that blood products injected into the pathway with a small-caliber needle before the biopsy or drainage instrument would create a “seroma” of the missing product (e.g., platelets, FFP). The concentration of the blood product would be “astronomical” in the local site. If the instrument encountered a vessel in the path, the local product would form a local clot. Because there was no other alternative for our radiologist colleague, we tried the method and it worked ( Fig. 67-30A ). With drainage of hemothorax, the patient improved symptomatically and lived for 3 months afterward.

We have since used local injection of platelets and FFP in patients who have an abnormal prothrombin time or low platelet count as a partial solution to the problem. An even more extraordinary case involved a teenager with hypersplenism whose low platelet count was unresponsive to replacement. To exclude the possibility of portal hypertension secondary to portal vein stenosis, a splenoportogram was ordered by the surgeons (see Fig. 67-30B ). Platelets were injected locally into the pathway during insertion (and removal) of an 18-gauge needle, and the study was performed without incident.

The many clinical circumstances under which we feel use of this technique is justified include: (1) the patient is too ill to wait for infusion therapy and in a life-threatening situation; (2) an existing condition of patient (e.g., pulmonary embolus, failed vena cava filter, recent coronary stent placement, cerebral occlusion) requires maintenance of anticoagulation; and (3) failure of the patient to respond or recent history of failure to respond to replacement therapy (i.e., antibodies, hypersplenism).

If these factors are present, we perform the local injection of blood elements as follows. After preparation of the site with povidone-iodine, the site and pathway are prepared in two steps. The syringe containing lidocaine is mixed 50% by volume with either FFP or platelets. The pathway is anesthetized (which includes blood product) in the routine manner, ensuring that the pathway is treated from skin to the organ surface and edge. Next we make a repeat injection of approximately 5 to 10 mL of blood product into the expected trajectory of the needle, including the skin, subcutaneous tissue, and surface of the organ and target (see Fig. 67-30 ). It is crucial that the entire pathway be injected, not simply the subcutaneous tissues or the capsule of the organ; bleeding can potentially occur from any site along the pathway and can push out any localized plug of clot. When a biopsy rather than a drainage is being performed, it is prudent to use a coaxial system that includes a guidance cannula and the biopsy instrument. The appropriate samples can be taken and the needle removed, leaving the cannula.

After the biopsy, the blood product is injected in the pathway through the coaxial cannula as it is removed ( Fig. 67-31 ). If bleeding occurs, one can inject more blood product into the site to provide hemostasis (see Fig. 67-31 ) (this has not happened to date in our experience), or one can use the coaxial/coil/thrombin method if bleeding continues (see subsequent technique and Fig. 67-32 ). If a drainage is being performed, no coaxial system is used. After this step, the drainage procedure is performed in the routine fashion. Our results over the past 5 years have been very favorable.

The rationale for this technique is that the local injection creates a local “seroma” of the missing product so that if any adjacent vessel is inadvertently punctured, the local elevated concentrate of the blood product provides material for an effective blood clot. With FFP the seroma concept is appropriate. With platelets, the platelets probably undergo immediate activation, which provides overabundant amounts of thromboplastin to initiate the thrombosis cascade.

From a physiologic perspective, investigators have shown that a native clot that forms at the site is superior to an injected “clot.” A clot formed after restoration of local or systemic coagulation factors is of better quality than one formed when an imbalance of coagulation factors exists. A “balanced” clot is more effective because studies have shown that the strength of a clot is due half to the platelet agglutination and half to the serum proteins.

Occlusive Hemostatic Methods for Cutting Needles.

The most consistent and practical method to prevent localized bleeding after a biopsy is placement of more solid occlusive material through a coaxial system after the biopsy has been taken and the needle has been removed. Four such materials have been used: Gelfoam, occlusive liquid “glues,” angiographic coils, and angiographic coils with thrombin.

In regards to injectable “glues,” numerous products have been evaluated in animal models and approved for clinical use, but in our experience their application is difficult. These products consist pri­marily of fiber material such as cellulose/Gelfoam or liquid protein material such as fibrin glue, which solidifies in the pathway after injection. A major problem of these products is that the fixation time of these materials is not predictable. When these are injected through a needle pathway, one cannot always be confident of the placement of the protein material. As the injection cannula is withdrawn, the protein plug may be pulled out. From the research data in the previous section, one appreciates that such artificial clots are not as good as natural clots.

Injection of Gelfoam through a coaxial system was reported by Chuang and colleagues. This method does not always work well because the Gelfoam is difficult to position properly and may be pulled out with the cannula, similar to the injectable materials.

Coaxial sheath with angiographic coils was reported by Allison and Adam. Although this technique has been reported as effective, we have altered it to use thrombin because angiographic coils have been changed from their original design. Until about 15 years ago, the coils had fabric on the end intended to serve as foci or anchors for clot formation. Those early coils had “wool” fibers on the coils, which was very thrombogenic, but the material was replaced with Dacron, which is not as thrombogenic. After being frustrated with the use of such coils because clot did not form quickly, we devised the coaxial/coil/thrombin method that we now prefer, which is the use of a coaxial system, angiographic coils, and thrombin.

We modified our technique to add thrombin after numerous studies. In a study of many coagulopathic animal experiments, we found that the coil with thrombin has three distinct advantages: (1) the coil is positioned precisely at the site of the biopsy “cut,” (2) the spring coil stays in place and does not “pull out” with the coaxial cannula, and (3) there is instant clot formation for the thrombin. After one becomes familiar with this technique, one easily prepares the items required for hemostasis either before a procedure or during the procedure when bleeding from the guidance cannula is noted. The coaxial cannula must always be left in place for access to insert the material. Preparation for this technique can be made before the procedure if the bolus injection shows increased vascularity or large collateral vessels. Required equipment includes two vascular dilators 6F or larger, angiographic coils, and thrombin ( Fig. 67-32A and B ).

Before the coaxial/coil/thrombin technique can be executed, the equipment must be assembled such that the dilators are loaded with the coils and thrombin so that they can be inserted as needed and noted in Figure 67-33 . The coils are loaded into the dilators and pushed carefully to the end of the dilator (see Fig. 67-32C ) and not beyond; holding the finger over the end as they are inserted prevents inadvertent deployment (see Fig. 67-32D ). The thrombin is mixed according to the instructions and injected into the dilator slowly while holding a finger loosely over the end (see Fig. 67-32F ). The idea is to permit the thrombin fluid to flow through, but not eject the coils if the injection is too forceful. Figure 67-32G shows a radiograph of a simulated biopsy, with the inner stylet of the needle inserted. Figure 67-32H shows a radiograph of the coaxial cannula after removal of the biopsy needle, and the coil with thrombin is pushed out the end of the cannula.

In cases in which an enhancing lesion is to be biopsied (see Fig. 67-33A ), we prepare the equipment as noted previously and inform the attending physician and the patient of the increased risk of bleeding and the method to be used to prevent bleeding. It is the prerogative of the patient and the physician to decline the procedure, but if it is decided to proceed, the routine steps are taken as noted in Figure 67-33B through D . The biopsy using the coaxial guidance system is performed in the routine fashion. The biopsy is performed through the cannula, and the tissue sample is removed with the cannula left in place (see Fig. 67-33B and C ). Through the cannula (see Fig. 67-33D ), an angiographic wire is used to push the coil through the dilator into the cannula and out the end of the cannula ( Fig. 67-33E ). After the procedure, the patient should lie quietly for 30 minutes to let a clot secure the occlusive material in place. If patients get up too quickly, the device may dislodge into the peritoneum if they happen to cough or exert themselves; there is no direct harm to the peritoneum except for possible bleeding from the site. This method may be used in the liver or other sites as needed. This coaxial method can be used with either a guided procedure of a mass or a diffuse sample of the liver.

Occasionally a unanticipated potential problem with bleeding may occur ( Fig. 67-34 ). In such cases, if excessive bleeding is observed from the hub of the cannula after biopsy of a lesion (see Fig. 67-34A and B ), one could decide to do nothing, but we prefer to proceed with this hemostatic method. If bleeding occurs, a stylet is inserted to delay the bleeding temporarily, and the equipment is prepared as noted earlier. If one has the equipment available, preparation takes only several minutes. As noted earlier, the coils are inserted through the cannula to provide hemostasis (see Fig. 67-34C through E ).

When first learning the technique, it is helpful to have fluoroscopy available because it is easier to follow the insertion of the thrombin/coil in “real time.” One can use the topogram or rapid repeat CT scans, however, to monitor the coil insertion. Such procedures can be performed at the bedside with CT or US, but this requires extensive experience because with this approach the procedure is essentially “blind.”

Comparison of Aspiration and Cutting Needles.

The superior sample quality from a core cutting needle is no longer disputed. Numerous authors have reconfirmed the advantage of diagnostic accuracy with little difference in complications in most organ systems. There is no longer any controversy, and virtually all workers concur that cutting needles are superior to aspiration needles.

Selection of Aspiration Needle Type and Caliber.

The selection of aspiration needles is straightforward because there is clear laboratory and clinical information that indicates the best type and caliber of FNA needle to use. These points were shown by historical articles and have been reconfirmed over the years without controversy. In the earliest laboratory study by Andriole and associates, our group showed that the best samples with the least amount of tissue artifact were obtained by the small-angle bevel needle tip (25- to 30-degree angle) and by the largest caliber needle ( Fig. 67-36 ). This was later confirmed by Dahnert and colleagues.

In the direct clinical comparisons, these results were reconfirmed. Pagani showed with aspiration biopsies of the liver that an 18-gauge aspiration needle was better than a 22-gauge needle. In the lung, there has been no definite benefit shown with large aspiration needles, and there was a larger incidence of hemoptysis shown by Khouri and colleagues. In the pancreas, Dickey and associates of our group showed the diagnostic return for a 20-gauge needle compared with a 22-gauge needle, with no difference in complications. Reporting on the prostate, more recently Norberg and coworkers showed increased diagnostic yield of the 0.9-mm needle (20 gauge) over the 0.8-mm (22 gauge) or 0.7-mm aspiration needles. From our experience and the numerous reports as noted earlier, we have consolidated the needle choices in our institution to use the 20-gauge FNA needle in all organs. It is suitable for lung and solid organs.

Selection of Cutting Needle Type and Caliber.

Many variations of cutting needles exist, so numerous factors should be considered in the decision making. The first question is whether to use an end cutting needle or a side cutting needle for soft tissue or bone. There seems to be little controversy from either the scientific perspective or the clinical perspective about the merits and shortcomings of these needles. Numerous authors, including Bateson and colleagues and Andriole and coworkers, have shown that the side cutting needle provides the best sample without crush artifact or fragmentation of the core. For bone, the appropriate needle is the end cutting needle unless the target is associated with soft tissue. The cylindrical end cutting needle is the strongest and does not have the risk of bending. The side cutting needle with the thin waist of the stylet can bend even if within the cannula.

The first decision to be made with cutting needles is whether to use a manual or automated type; most operators choose an automated variety. A survey by Mayoral and Lewis showed that more than 90% of radiologists doing liver biopsies use these devices; the only variables are the vendor and caliber of the needle.

Probably the most important decision to be made is whether to use a single- or double-action needle. The terminology does not refer to the action of the needle, but rather the requirement of the operator. One can understand the two-step mechanism by reviewing the images of the manual type of needle shown in Figure 67-37 . A single-action device means that the operator needs to do only one thing: “push” the button or “pull” the trigger. The double-action device requires that one insert the inner stylet with the tissue receptacle and actuate the cutting action of the outer cannula as a secondary step. When the stylet is introduced, it defines the site of the biopsy. The double-action type is safer than the single-action device when there is critical anatomy close by.

In our opinion and experience, single-action devices should be avoided ( Fig. 67-38 ) for reasons described subsequently. When a small lesion in a precarious location is being sampled, one has to be absolutely certain about where the cutting action occurs. An experienced interventionalist knows that if a patient shifts position slightly, the internal anatomy can change by numerous millimeters or centimeters. This change can happen very subtly, and if undetected, catastrophe can occur if one is using a single-action device (see Fig. 67-38 ). In a case in Ohio that was litigated, a patient died because the main pulmonary artery was biopsied inadvertently. Because a single-action needle was used, the final position of the stylet was not seen because it fired instantly before the cutting action. With the double-action needle, the stylet is inserted separately, and the exact location can be noted. The cutting action occurs precisely at that site. Hemostasis is more effective with use of a hemostatic occlusive method such as the coaxial/coil/thrombin technique.

The final decision to be made is the caliber of the needle. This decision depends on the amount of tissue required for the histopathologic studies to be performed and the organ sampled. In the new era of tissue biomarkers, routine and possibly repetitive biopsies are expected to become the norm. Workers from our group have provided some insights relative to needle caliber and its relationship to bleeding and the amount of tissue acquired. Gazelle and colleagues showed that increased bleeding occurs with certain calibers of needles in the normal and coagulopathic states (see Table 67-1 ). From the work by Plecha and associates, one can determine the amount of tissue obtained with each size of needle ( Table 67-2 ). Plecha's group studied the outcome in an animal model using 20-gauge, 18-gauge, and 14-gauge needles. These authors evaluated the amount of diagnostic information compared with the amount of bleeding by weighing the amount of DNA (nuclear material) and blood in the tissue sample. They noted that the larger the amount of tissue required for the diagnostic information, the more efficient it was to use a large-caliber needle.

Equally important as obtaining adequate samples is reducing the possibility of bleeding. Recognizing that bleeding is the result of multiple factors, including the natural vascularity of the organ (spleen > kidney > liver), the specific vascularity of the tumor, and the random event of striking a vein or artery, one can choose the optimal caliber. Smaller needles require more passes, producing greater statistical chance of striking a vessel. We select the caliber of the needle according to the type of organ being biopsied, the specific type of histopathologic evaluation to be performed, and the characteristics of the organ being sampled (see discussions under each organ system).

Assessing vascularity before sampling an abnormality is especially important with a cutting-type needle. In most instances, the vascular character of the lesion is apparent from a prior contrast-enhanced study. If not, one should perform a study with a bolus of contrast material and repetitive scans at the site of the biopsy to evaluate the vascularity. Not only do cancers have increased vascularity, but unusual benign vascular abnormalities (e.g., arteriovenous malformations [AVMs], aneurysms, cavernous transformation of the veins) occur in virtually all organs; detection of such lesions by a dynamic scan prevents a bleeding complication.

From our experience and review of the literature, the vascular nature of an abnormality is a major determinant of bleeding complications. High bleeding rates occur even with 22-gauge needles when vascular tumors such as hepatomas are biopsied. In our latest series of 200 liver biopsies, the complication rate with a 14-gauge needle was 0.5% because we assessed tumor vascularity with a dynamic bolus scan over the site of the selected target. The lesions that were too vascular or close to vessels were deferred; this study was before our refinement of the coaxial/coil/thrombin technique. It is likely that we would not have deferred any biopsy because of vascularity had we had the technique during this period.

As noted earlier, with coagulopathic patients, the smallest caliber appropriate should be used depending on the “contingency” hemostatic method used—systemic blood products, local injection of blood products, or the coaxial/coil/thrombin technique (see “Coagulopathic Treatment”). The systemic replacement to normal status permits any size of needle, as does the local injection method. The coaxial/coil/thrombin technique requires a coaxial needle system of 18 gauge or larger; we defer to the 18 gauge.

Numerous clinical articles have studied the histologic information obtained from tissue samples for kidney and other organs. Hopper and colleagues showed that sufficient diagnostic tissue could be obtained from the kidney with an 18-gauge needle in pediatric patients. Haaga and associates found in a comparison of 18-gauge and 14-gauge needles that the diagnosis was equivalent for both devices. A survey of gastroenterologists and radiologists showed that most physicians use an automated side cutting needle in the range of 18 gauge.

Although the use of the 18-gauge needle is the norm in the United States, there is good reason for radiologists to begin using large-caliber needles in selected cases. In recent years, new methods developed at our institution in conjunction with the National Cancer Institute portend to enhance the development of new cancer agents and drugs by using samples from 14-gauge needles in selected cases. As reported by Dowlati and coworkers, we had a 0.5% bleeding rate with 14-gauge cutting needles in the liver in a series of 200 patients. In these trials, large needles were used to recover histologic samples, suitable for microchemical assays of molecules that were the specific target of cytolytic or modulating agents. These trials permitted more rapid drug development at lower costs. Improvements in determination of pharmacokinetics and dosage regimens occurred to the extent that a new term, the BMD (biochemical modulatory dose), was coined. In some cases, this term was substituted for the MTD (maximum tolerable dose before unacceptable complications occur).

Possible overuse of cutting needles.

Having innovated cutting needle biopsies using CT guidance and advocated their use for many years, it is ironic that we are now concerned that cutting needles occasionally are used too frequently by some operators. We have always been strong advocates of their use but continue to emphasize the importance of judgment when selecting them to achieve the greatest tissue yield, using proper patient selection, approach, and techniques to minimize the potential complications. We have not changed our opinion on the superiority of the tissue samples, but rather we are concerned about the cavalier attitude some operators have for minimizing the possibility for complications.

We have noted that some physicians use the cutting needle with impunity and put patients at great risk and themselves in great legal peril. Lest one misconstrue these statements as indicating that we have a high complication rate, our complication rate (according to the National Cancer Institute, which monitored all of the NIH studies and reviewed the charts) was 0.5% in a group of 200 patients (see “ Liver ” section).

Figure 67-39 presents a case that shows the exact opposite of all our recommendations. This case was performed by a community radiologist in a nonuniversity hospital and resulted in two significant life-threatening complications. The radiologist used an 18-gauge cutting needle for a deep lung lesion, which predisposed to severe hemoptysis (a 20-gauge cutting needle should be used for such deep lesions). The cutting needle was deployed with the inner stylet outside of the mass and adjacent to a vessel (see Fig. 67-39 ). The patient had a large pneumothorax and large hemorrhage in the right lung; the pathologic diagnosis was inconclusive (not a surprise, seeing that the needle was not properly positioned in the mass).

The reader is urged to take our recommendations and suggestions because we have performed these procedures for 33 years, including approximately 20,000 personal experiences. Everyone must develop their own guidelines, but it is always better to err on the safe side.

Axial Slice.

Positioning a needle in the same xy plane of an abnormality is the simplest approach if possible. In these cases, with the needle in the superficial tissues, the angle of the needle is adjusted with repetitive scans until angled correctly. When the correct angle is achieved, the device is inserted to the desired depth either with incremental adjustments and repeat scans or to a premeasured distance. Subtle adjustments of the needle can be made using the deflection principles, or cannula manipulations noted in the following sections may be used meticulously to optimize the needle position.

Localization of the Needle Tip.

After insertion of the needle, one occasionally may have some difficulty in localizing the tip. Several methods can be used to solve this problem. The simplest method is to take a longitudinal scan and take an axial slice at the end; this is virtually foolproof and is the most expeditious. Another method that can be used if the angle of the needle is slight involves taking multiple scans in one direction until the end of the needle is reached ( Fig. 67-43 ). Some new scanners have a cluster scan feature consisting of three or more scans, which is well suited for this purpose. When the end of the needle is found, it appears to have a square end and may have an artifact from the end. Occasionally the end of the needle is not square but is easily appreciated by noting when the end is no longer seen on sequential scans.

Unintended Needle Deflection.

With a bevel needle, the needle either moves straight or deflects depending on the hardness of the mass, angle of the needle, and speed of insertion. A hard mass is likely to deflect the needle if the trajectory pathway is not perpendicular to the tangent of the entrance point. If a beveled needle is pushed slowly, it is more likely to move the point away from the flatter part of the needle. If a needle is pushed quickly, it tends to go straighter. For very hard masses, deflection may not be avoided unless one continuously rotates the needle during insertion or a stiff guidance cannula greater than 18 gauge is used. In moderate-density material, controlled deflection is possible.

Intentional Needle Deflection.

With the intentional needle deflection method, one can change direction of a needle when it is located in a patient below the surface of the skin. Purposeful deflection can change the angle of the needle at the latest stage of adjustment before striking the mass. This deflection is best accomplished by rotating the needle so that the flat side of the bevel is facing the area one wishes to avoid, and the point is directed more toward the target direction. The skin adjacent to the needle is pulled away from the point. This changes the fulcrum of the needle so that it is more angled toward the target ( Fig. 67-44 ). The needle is then pushed slowly. By using this method, one can change the direction of the needle by 10 to 15 degrees, which can amount to a 10-mm change; this can prevent the need for a major readjustment or reinsertion of the needle ( Fig. 67-45 ). This technique works only with a needle with a single bevel at the tip and not with needles with two bevels or a central point, such as a pencil or trocar point.

This technique also can be used with the thin stylet of a cutting needle ( Fig. 67-46 ), which permits repositioning without removing the entire needle. To achieve this repositioning with the double-action technique, the needle is positioned adjacent to the target, slight pressure is used to deflect the shaft of the needle away from the bevel, and the stylet with the bevel is pushed slowly into the target.

Coaxial Cannula Manipulation.

Because the coaxial approach has such benefits as described earlier, one should consider the possibilities with the guidance cannula. Because the guidance cannula is stiffer than the basic needle, it can be used to torque the needle. This is done by simply angling the cannula in the desired direction. One must be aware that such devices are stiff, and moving such a device potentially can produce shearing.

The intended deflection noted earlier can occur with a coaxial cannula. Another technique that can be used to move a needle into a difficult area combines the benefits of a Chiba needle with a cannula. For lack of a better term, we have called this a cannula slide maneuver . If one wishes to use a cutting needle beyond a vital structure, one can advance the cannula in the following fashion. Through the cannula, one pushes a bevel Chiba needle beyond the vessel. Then one advances the cannula forward over the small-caliber needle. When the cannula is distal to the vital structure, one can use the cutting needle safely. We have used this technique in the mediastinum and the pelvis (see Figs. 67-77 and 67-131 ) in lung sections.

Rotation of Cutting Needle to Optimize Sampling.

In some cases when a biopsy is being attempted with a side cutting needle, the placement is not completely optimal because the “needle gap” misses the mass slightly. In such cases, one can increase the probability of getting an adequate sample by rotating the needle so that the gap is oriented toward the center of the mass ( Figs. 67-47 and 67-48 ).

Freehand Method.

With the freehand approach, it is best to consider the approach to the lesion as a 3D system. The final vector or angle of the needle to the lesion would be a combined angle consisting of the xy angle and the yz angle. When doing this, it is best to establish the xy angle in the superficial tissues of the lowest slice before attempting the z angulation. When the xy angle is proper, one can consciously hold that angle constant while moving the needle into the z axis ( Fig. 67-49 ).

When the needle is close to the target site, numerous scans are made over the path of the needle to judge the position in 3D. If a minor adjustment needs to be made, one should use the tip deflection method described earlier under “Intentional Needle Deflection.” Using this technique after the first major angulation has been accomplished saves time and makes success more likely.

Angle-Estimate Method.

The angle-estimate method is a permutation of the freehand method, but more care is given to trying to calculate or simulate the angle more precisely. Most simply, this is done by using a calculation of the Pythagorean angle from the scans to arrive at a numerical angle in the different planes ( Fig. 67-50 ). The execution of the instrument insertion is performed by the freehand method or by using a manual angulation device.

Patient Positional Changes.

Placing the patient in lateral or oblique positions produces shift of internal anatomy so that loops of bowel or organs may move sufficiently to open a procedure trajectory pathway. In most instances, bowel shifts to “float,” and solid organs “gravitate” or “roll” depending on capsular attachments. Although one might predict such movements, trial and error may be needed. If one positional change does not work, rotation to the oblique or the opposite side might produce the desired effect.

Injection of Gas or Fluid.

Although one can penetrate certain loops of bowel and uninvolved organs in some circumstances, it may be advisable to avoid certain structures or to protect them from unintended damage during ablation procedures. Two methods have been used for dealing with this problem: injection of carbon dioxide and injection of saline.

We reported in 1986 that carbon dioxide insufflation or room air can be used to move structures short distances. Other authors have proposed the use of injected saline to increase the size of pathways through which one can insert a needle. This technique has been used predominantly in the chest for widening the paraspinal space, but potentially it could be used in other locations (see later under “ Lung and Mediastinum ”) ( Fig. 67-51 ). Langen and colleagues studied the efficiency of carbon dioxide and saline in an animal model to move structures from a needle trajectory pathway. With equivalent amounts of gas or saline, they found that the saline produced substantially more movement of organs than carbon dioxide. Neither fluid nor gas moved solid organs, such as kidneys.

More recently, many authors have reported using such materials to prevent inadvertent damage to adjacent structures during ablation procedures. Authors have reported the use of saline or water to provide a conductive layer between the thermal ablation site and any adjacent normal structures. Laeseke and coworkers reported the use of dextrose water and saline to prevent thermal damage to adjacent uninvolved structures. In their animal model, dextrose prevented damage better than saline, permitting smaller sites of damage (perhaps because of conductivity of saline). Kariya and associates reported the use of injected carbon dioxide to produce “insulation” between the thermal site and normal structures. The maximum distance was slightly greater than 1 cm. Caveats for saline injection include patient discomfort, decrease in target lesion conspicuity, and adverse effects of absorbed fluid in patients with kidney or heart disease.

Both of these methods have a similar problem. Depending on the location of connective tissue, either the gas or the fluid may dissect cephalad or caudad and not produce substantial enlargement in the axial plane. Nevertheless, both of the methods can work in selected cases. Another caveat is the effect of severe inflammation or tumor infiltration. If either of these is present, movement or displacement may be impaired.

The procedure is performed easily. After the trajectory is planned, the operator decides from the scans the proper location of the gas or liquid. The operator inserts a needle and injects by hand 50 to 100 mL of the material ( Fig. 67-52 ). If the intended loop of structure does not move, the operator attempts to reposition the needle or repeats the injection. Depending on the length of the procedure, repetitive injections might be necessary because the material disperses gradually along fascial planes, and the displacement decreases accordingly.

External Manipulation of Bowel Loops.

In some patients, it may be possible to move an intervening loop of bowel by manual palpation through the skin. Anyone who has ever performed fluoroscopy is familiar with how mobile loops of bowel may be. Tuncali and colleagues used manual displacement of bowel to provide greater clearance between the iceball of cryoablation and adjacent bowel loops to prevent collateral damage. In a group of 14 patients, mean distance between tumor margin and closest adjacent bowel increased from 0.8 cm (range 0-2 cm) before external manual compression to 2.6 cm (range 1.6-4.1 cm) with manual displacement ( P < .01). Occasionally we have used this technique for biopsies, and it has been beneficial.

CT.

CT is the modality of choice for the biopsy of most lung lesions regardless of their locations. Because the shape and angle of the ribs varies from the front to the back of the ribs, approaches for biopsying different lesions must be varied ( Fig. 67-53 ), as noted in later sections. Lesions that were problematic for fluoroscopy, such as lesions in the apices ( Fig. 67-54 ; see Fig. 67-35 ), costophrenic angles ( Fig. 67-55 ), costovertebral angles ( Fig. 67-56 ), pleurae ( Fig. 67-57 ), and hila, and lesions obscured by fluid and mediastinum, are well suited for CT. Parenchymal lesions with ill-defined edges are well seen, and one can distinguish necrotic areas or contiguous areas of infiltrate adjacent to the solid portions of a mass.

Avoidance of complications is improved because visualization of anatomy is excellent, and one can exploit anatomic detail to avoid problems in patients with parenchymal disease or vascular anomalies or blebs. In such cases, if one takes thin scan sections, one can sometimes find a pleural attachment that permits a “risk-free” access, avoiding penetration of the lung ( Fig. 67-58 ). In patients with an unknown or unexpected vascular anomaly, bolus dynamic CT scan is an important adjunct for the mediastinum and hila because major vessels can be opacified and delineated, making separation from masses in these areas easy.

PET Scanning.

PET scanning has become a valuable tool for the diagnosis and characterization of lung masses and is commonly a prime indicator for biopsy. Cancer is hypermetabolic and consumes larger amounts of radioactive glucose than benign or noncancerous masses ( Fig. 67-59 ). In most cases, PET is unparalleled for characterizing the cancerous nature of pulmonary nodules. Its use has revolutionized the practice of medicine relative to pulmonary nodules.

Occasionally an inflammatory site or nodules in the lung may be problematic because of associated inflammation. Although delayed scans have been used in the PET protocol to reduce the uptake and differentiate cancer from inflammation, this has not always been effective. After encountering numerous cases intended for percutaneous biopsy, we used a new technique to reduce the inflammatory reaction to eliminate the question of tumor in such cases. In nine patients referred for biopsy, we used a new technique that is not yet FDA approved. For problematic cases, this new technique can reduce the inflammatory vascularity and FDG uptake in inflammatory foci sufficiently to eliminate some false-positive findings for inflammatory pulmonary nodules ( Fig. 67-60 ).

Target and Trajectory Planning.

A diagnostic series of scans is taken in anticipation of the procedure. From the current scans, the mass referred for biopsy is reevaluated. If the lesion is stable or larger, the procedure should proceed. If the lesion is significantly smaller, one should consider deferring the procedure until a later date after follow-up scans are performed.

One selects the tentative target for biopsy and the appropriate pathway for the trajectory back to the skin surface. If the mass is large, one selects a portion of the mass that is not necrotic and a portion that is definitely solid if there is associated infiltrate or collapse. When there are questions about viability of the mass, correlation with a PET scan may be helpful.

The trajectory pathway to be used (which influences the entrance point selection) should avoid bullae or blebs, bronchi, and vessels. Penetration of bullae or blebs may cause pneumothorax. Puncture of the bronchi causes coughing, which can cause pneumothorax or possible air embolism to the vessels. If possible, one should choose the shortest pathway to the lesion, realizing that the distance through aerated lung is directly related to the pneumothorax rate (see “ Lung and Mediastinum ,” later). If there is an infiltrate or pneumonia, one should plan to traverse those areas if possible because this lessens the chance of pneumothorax. If there is a fibrous attachment or pleural thickening nearby, the trajectory should traverse them; it is likely that the fibrotic tissue anchors the visceral pleura to the parietal pleura, lessening the chance of a pneumothorax.

Bolus dynamic contrast scan over the chosen target site should be performed for numerous reasons. First, if the patient has any history of bruit on auscultation of the chest or if the scan appears to show a prominent vessel adjacent to or entering the mass, a bolus scan probably should be performed ( Figs. 67-61 and 67-62 ). The other circumstance would be in instances when there is no clear history of smoking or other risk factors for a tumor. Most importantly, bolus scanning is absolutely indicated if any consideration of a cutting needle is being made. The short time required for such a study is a small price to pay for avoiding a life-threatening catastrophe.

Patient position may be adjusted depending on the target site and the specific anatomy of the trajectory. In most instances, it is simplest to have the patient lie in the prone or supine position, but other positions also might be considered ( Fig. 67-63 ; see Fig. 67-35 ). In some cases, changing the patient position may “clear” the pathway for the biopsy. Specifically, the position of the scapula and the relationship of the ribs to the mass may change with an oblique position, converting an unlikely approach to a suitable one (see Fig. 67-63 ).

If the lesion appears mobile, which is common close to the diaphragm, the patient can be positioned with the ipsilateral side down, which may lessen motion of the diaphragm (see Fig. 67-35 ). If the lesion is small and moves with respiratory motion, it may be wise to put the patient in an ipsilateral decubitus position and insert the needle through a direct anterior or posterior approach. This position minimizes the motion of the ipsilateral diaphragm because of the weight of the abdominal organs against the diaphragm. This restriction of motion minimizes movement of the lesion and simplifies the performance of the procedure. Kinoshita and colleagues and Rozenblit and associates reported that if a biopsy can be done with the ipsilateral side dependent, the pneumothorax rate is reduced.

Procedure Evaluation and Planning.

During the procedure, vital signs, neurologic status, and pulse oximetry should be monitored. Analgesics and sedatives can be administered as needed (see earlier under “ Patient Cooperation and Sedation ”).

Instrument choice and techniques.

Our needles of choice for lung and mediastinal biopsies, considering the diagnostic outcome and complication rates as noted in those sections, are (1) the 20-gauge FNA needle (spinal needle with 30-degree bevel is equivalent) for aspiration biopsy, (2) 20-gauge cutting needle for suitable parenchymal or pleural masses, and (3) 18-gauge or larger cutting needles for larger pleural-based masses. A single-needle method or a coaxial method can be used for a needle aspiration of the lung. With the single needle, the needle is started into the superficial tissues and a scan is obtained to verify its angulation. When the correct angle is determined, the needle is inserted to the correct depth. The coaxial method should be used for deep or small lesions. The guidance cannula is inserted into the chest wall and directed immediately adjacent to the lesion. The needle is inserted through the cannula into the mass.

When a mass is accessible and in the appropriate location, a cutting needle may be used judiciously. The most critical factor with any cutting needle procedure is that a bolus injection be performed before the procedure. Although infrequent, vascular lesions, such as AVMs or vascular sarcomas, may occur, and catastrophe can be avoided by aborting the procedure (see Figs. 67-61 and 67-62 ).

If a lesion within the parenchyma is accessible and of sufficient size, a 20-gauge cutting needle may be considered because the risk is no different than an aspiration needle (see Fig. 67-35 ). The size and location of the mass must be adequate to permit complete placement of the cutting area within the lesion ( Fig. 67-64 ). In such cases, the intervening lung has no additional trauma than an aspiration needle because there is no caliber difference or exposure of the traumatic tip to the lung.

One must be prepared to compensate for deflection and bending of the needle when the stylet is inserted. Benign lesions tend to be very hard, and the surrounding lung does not provide much resistance. Such lesions may bend the stylet or deflect it to the side ( Fig. 67-65 ).

Postprocedure Care.

Postprocedure care is simple and consists of observing patients for severe hemoptysis and pneumothorax. Immediately after completion of the procedure, I prefer to have the patient lie still for 5 to 10 minutes to let the puncture site seal. Moore and coworkers have shown that the incidence of pneumothorax is reduced if the patient is positioned with the puncture site down after the procedure. The pneumothorax rate for patients placed in this position was 17.9%, whereas the rate was 33.6% for patients not so positioned. Chest tube placement rate was 9.8% for patients who were positioned with the procedure side down compared with 0.4% for patients who were not.

Before removing the patient from the table, we take a repeat scan to evaluate for postprocedure problems. Although Miller and associates reported that CT and plain radiography were equivalent for pneumothorax detection, Bungay and colleagues reported that CT is more sensitive for detection of pneumothorax (radiography detected only 22 of the 35 pneumothoraces detected by CT). A small pneumothorax is of little significance, as noted later. Some bleeding in the parenchyma in the area of the needle track occurs frequently, but it is of no significance. If there is any question of the patient's stability, oxygen should be administered, and appropriate treatment should be given.

Postprocedure chest films should be taken 1 hour, 4 hours, and 8 hours after the procedure. It makes no sense to use valuable CT time for follow-up of pneumothorax; chest tube insertion is required only with larger pneumothoraces when patients are symptomatic. Perlmutt and coworkers showed that pneumothoraces could be detected in the following percentages at the noted timed delays: 89% immediately after procedure, 9% after 1 hour, and only 2% after 4 hours.

Accuracy of Diagnostic Tissue Recovery.

The success rate of lung biopsies has varied widely among the many reports of fluoroscopic and CT-guided procedures and with the use of aspiration and cutting needles. Before cutting needle usage, favorable results were reported for aspiration cytologic results. With increased use of cutting needles, the literature shows improved recovery of diagnostic tissues for all malignant and benign lesions.

Pneumothoraces—Factors Affecting Incidence.

Effective administration of local anesthesia through the trajectory site to the pleura helps reduce pneumothoraces. If the patient does not have the pleura adequately anesthetized, even the most cooperative patient may jerk at the instant a needle traverses the pleura, potentially tearing a small hole and producing an air leak.

Theoretically, placing a postbiopsy patient with the ipsilateral side dependent minimizes the incidence of a pneumothorax. This theory is controversial and not yet confirmed because there are conflicting reports on the topic. In a series of 262 patients, Moore and colleagues reported that the pneumothorax rate was reduced in a statistically significant manner. Collings and associates reported in a series of 423 patients that no benefit occurred.

Although early authors attributed the high pneumothorax rate to the prolonged length of the procedure, a more recent report by Ko and coworkers refutes this claim. These authors found no difference between patients with procedures with different times. Two major determinants of pneumothorax rate are the status of the lung and the number of needle passes. If a patient's lung has severe chronic obstructive pulmonary disease (COPD), more blebs are present. Decreased lung compliance in patients with interstitial disease favors collapse of the lung when a small air leak occurs.

In prior editions, we noted from our experience numerous parameters important for pathway selection, which have since been validated by reports in the literature. One can minimize the possibility of a complication by looking carefully for a safe pathway through the lung to the nodule being biopsied. In such cases, one might locate the attachment to the pleura of an area of fibrosis or thickened pleura that can be traversed by the needle. By planning the trajectory through such an area, only a minimal amount of parenchyma is traversed or the fibrosis can serve to keep the layers of the pleura attached and avoid a pneumothorax ( Fig. 67-67 ). Ko and coworkers reported that pleural thickening associated with prior surgery decreased the pneumothorax rate ( P < .05). By looking for a small amount of “fat” that may be tented at the pleura, one usually can distinguish between a mass that is simply adjacent to the pleura and one that has a fibrous extension. This type of approach is especially important in high-risk patients with pulmonary hypertension, severe restrictive disease, or other serious underlying pulmonary problems.

Minimizing the pathway of lung crossed ( Fig. 67-68 ) and using a pathway through nonaerated lung can minimize the incidence of pneumothorax. The significance of minimizing the needle pathway length was first evaluated by one of our colleagues, Locke (Locke J, Haaga JR: Personal communication, unpublished data), who studied a series of 235 cases from our center and looked at the incidence of pneumothorax compared with the distance of the lung traversed. The distances crossed and the corresponding incidence of symptomatic pneumothorax were 1 cm, 3.7%; 1 to 2 cm, 10.7%; 3 to 4 cm, 19%; and 5 to 6 cm, 44.4%. Because these measurements were taken from CT scans, they were accurate without errors in magnification; other authors who have not supported this relationship made their measurements from radiographs, which are more subject to error. In recent years, other authors have confirmed that lesion depth was the prime determinant of pneumothorax for CT procedures.

That traversing nonaerated lung, as we initially published in prior editions, does diminish pneumothorax has been documented. Haramati and Austin showed convincingly that if aerated lung was avoided, the pneumothorax rate decreased significantly. These investigators found that the pneumothorax incidence through aerated lung was 46% compared with 0% through nonaerated lung. Cox and associates reported 15% pneumothorax for nonaerated lung and 50% for aerated lung. Stated differently, if there is an area of infiltrate adjacent to a nodule, one should intentionally traverse the consolidated area ( Fig. 67-69 ).

Status of the lung was evaluated by Fish and colleagues and Miller and associates. Topal and Ediz and Cox and colleagues showed a definite relationship of pneumothorax to lung disease. Fish and colleagues evaluated chest films and spirometry to determine the presence of COPD, and they found that patients without COPD had a pneumothorax rate lower than patients with COPD. With spirometry measurements, the ratio for pneumothorax was 45% for COPD patients compared with 25% for patients without COPD. Using the chest x-ray as the standard, the rate was 42% for COPD patients compared with 25% for patients without COPD (see earlier under “ Pneumothoraces—Factors Affecting Incidence ”). Cox and colleagues reported that patients with emphysema were three times more likely to have a pneumothorax. Avoiding cystic abnormalities reduces the chance of a pneumothorax (see Fig. 67-55B ).

Several authors have reported a blood patch technique to prevent pneumothorax. With this method, a coaxial needle method is used, and a blood clot is injected through the long cannula during its removal from the lung. The results of this method vary; Bourgouin and associates reported no difference in outcome when it was evaluated in a randomized prospective fashion. Lang and coworkers studied a group using a blood patch method; in patients with emphysema, pneumothorax occurred in 3 of the 20 patients (15%) who received autologous blood clot and 10 of the 14 (71%) patients who did not ( P < .001). In our opinion, the difference in results is probably based on the exact method of instilling the blood clot. If clot is simply deposited at the pleural margin, it is unlikely to seal the leak of air. If blood is injected throughout the pathway of the needle within the lung and the pleural margin, a seal is more likely to succeed.

Hemoptysis—Factors Affecting Incidence.

Hemoptysis is expectoration of blood after biopsy; it does not refer to the imaging appearance of a small amount of bleeding, Commonly, minimal localized hemorrhage in the parenchyma may occur after a biopsy, but it is of no significance ( Fig. 67-70 ). The factors affecting hemoptysis are the needle caliber, depth of lesions, coagulation status, presence of congestive failure, and an improper trajectory that passes through a moderate-sized pulmonary vessel. Excessive damage to the lung parenchyma with a cutting needle also can occur if the needle is not positioned properly (see Fig. 67-39 ). These circumstances should be avoided if possible. Although this is an unsettling event for the patient and the physician, hemoptysis was common in the early days of fluoroscopic biopsy. Needle caliber directly affects hemoptysis; this was confirmed by Khouri and coworkers, who noted in their series that the hemoptysis rate was 21% with an 18-gauge needle but only 5% with a 20-gauge needle.

Deep lesions and hilar lesions are at greatest risk because of the long pathway and proximity of blood vessels. Yeow and colleagues reported depth and size of lesion as the most important factors for hemoptysis. The cases reported in the literature were associated with ill-defined masses adjacent to the hila. Also, one must remember the importance of bolus studies in selected patients (see Figs. 67-61 and 67-62 ). Vascular lesions, aneurysms, and AVMs are very vascular and may bleed. Bolus dynamic scan is crucial to avoid biopsy of such lesions.

Patients with coagulopathies, pulmonary hypertension, or congestive heart failure are at greatest risk for hemoptysis and intrabronchial bleeding (as noted later under “Intrabronchial Bleeding”) and should not undergo biopsy without correction of the basic problem. Biopsy in a patient with congestive heart failure should be avoided because the pulmonary vessels have elevated pressure, are edematous, and are more fragile, being predisposed to tearing.

Anticoagulated patients should never undergo biopsy, even if the anticoagulation consists only of aspirin. The lung parenchyma has a very low tissue pressure because of the alveoli, and there is no compressive effect to help retard bleeding.

Fatal Events.

We have located nine reported deaths in the literature from percutaneous lung biopsy, five from endobronchial bleeding and four from air embolism. Having read those reports, some valid observations can be made.

Hilar Lesions.

Hilar lesions are best sampled by bronchoscopy, and percutaneous biopsy should be employed only as a second choice. With percutaneous procedures, the length of the lung to be crossed and the proximity of vessels make the incidence of pneumothorax and hemoptysis more likely. The success rate for bronchoscopy is better and the complication rate lower than for percutaneous procedures. If a bronchoscopy has been attempted unsuccessfully, a percutaneous procedure is performed with the following guidelines.

Before the procedure, intense contrast enhancement of the vasculature should be performed to highlight the location of the anterior, lateral, or posterior approach to be taken. With such lesions, a coaxial short cannula method should be used. This method is especially helpful because, as noted earlier, before the biopsy needle is inserted, the guidance cannula predetermines the angle. With a long distance from the pleura to the hilum, a few degrees in error would result in a large localization error centrally.

Apical Lesions.

Careful planning for apical lesions is important because of the unique anatomy of the upper chest. Anteriorly the brachiocephalic vessels, mediastinal vessels, clavicle, and axillary vessels make the approach more difficult. Posteriorly the scapula covers a substantial portion of the posterior chest. Avoidance is best accomplished by having the patient fully abduct the shoulder so the scapula moves laterally. Careful palpation usually reveals the margin of the scapula. Another anatomic factor making the approach more difficult is that the ribs become smaller as one progresses from the midchest cephalad to the apex. It is usually easier to plan the entrance site so that the angle also is slightly caudal, which makes penetration of the intercostal space easier (see Figs. 67-35, 67-53, and 67-54 ).

Several final points should be made about unusual masses in the apex. Neurogenic tumors and lateral meningocele may occur in the posterior mediastinum adjacent to the neural foramina ( Fig. 67-71 ). Neurogenic tumors are hard and painful in biopsy ; a diagnostic sample seldom can be obtained. Puncture of a lateral meningocele yields clear fluid consistent with spinal fluid.

Pleural Lesions.

Pleural lesion biopsies are difficult to perform fluoroscopically because as the lesion is turned perpendicularly to the fluoroscope, it becomes en face and more difficult to see. With CT, sampling of such lesions is easy and provides considerable flexibility for needle selection. An aspiration needle can be used at any time; in cases in which it is indicated, a cutting needle can be used for a more adequate tissue sample. Goralnik and colleagues reported that these cutting needles could be used safely for pleural lesions. These investigators recommended their use when either lymphomas or unusual tumors such as mesotheliomas were suspected. An automated biopsy device also can be used if the lesion is large enough to accommodate the length of the needle action. Scott and coworkers noted that “by combining findings at biopsy and at CT (presence or absence of pleural thickness >1 cm, mediastinal-circumferential involvement, irregular contour), sensitivity and negative predictive values reached 100% and 1.0.” Maskell and associates reported that guided biopsy of the pleura is superior to unguided pleural biopsy. Adams and Gleeson reported on pleural biopsy of patients with suspected malignant effusions. In their series, accuracy was 91% overall and 93% for mesothelioma.

Several authors have reported on the use of long versus short pathways for biopsy of small subpleural nodules. For small subpleural nodules about 1 cm, Gupta and colleagues compared the usefulness of using a direct short pathway with a long pathway (entrance site and trajectory slightly away from the nodule). They reported better accuracy for a diagnostic result but a higher incidence of pneumothorax for the long pathway ( Fig. 67-72 ). The diagnostic yield in the short pathway group was 71% compared with 91% for the long pathway. Although the frequency of all pneumothoraces for both groups was the same, the chest tube rate was higher for the long pathway (38% vs. 17%). Ko and coworkers reported that a shallow angle of insertion was more likely to produce a pneumothorax.

Cavitated Nodules.

Cavitated nodules are slightly problematic because two major issues arise. To obtain viable diagnostic tissue, one must sample the wall carefully (see Fig. 67-72 ). If the needle enters the cavity, a pneumothorax is likely and there may be spillage of material into the pleural space; spillage of neoplastic or infectious material into the space should be avoided.

Mediastinal Lesions.

Before the use of CT, only a few authors ventured to perform biopsy of the mediastinum using fluoroscopic guidance. With the refinement of CT techniques, however, CT is now the modality of choice and is commonly used for this area.

Several limited studies by Lauby and colleagues, Jereb and Us-Krasovec, and Rosenberger and Adler showed the feasibility of mediastinal biopsies with fine needles guided by fluoroscopy. Weisbrod and colleagues and Westcott confirmed the benefits of fluoroscopic procedures but also noted some of the disadvantages and potential hazards. The most extensive series, consisting of 100 patients, was reported by Westcott. Although the results showed 96% accuracy, there were 10 patients with vascular aneurysms that created some difficulty. Five of the patients were excluded from a biopsy because of a difficulty in evaluating the mediastinum. In another five cases, inadvertent puncture of aortic aneurysms occurred; in one patient, a hemomediastinum developed.

Weisbrod and colleagues also reported several important observations. In their series they noted that although aspiration needles were satisfactory for metastatic disease, the needles were suboptimal for lymphomas. They also reported one case of cardiac tamponade that was successfully treated.

Mediastinal anatomy and approaches.

The mediastinum is an exciting area for the application of procedures. Before the advent of CT, when most biopsies were performed with fluoroscopy, this area was considered off limits because there was no reliable way to image the anatomy, either diagnostically or for intervention. With the evolution of CT imaging and its corresponding procedures, radiologists and clinicians have become confident of mediastinal anatomy and pathology.

The reliability of anatomic display and refinement of the interventional techniques are such that procedures of the mediastinum are now commonplace. These procedures require considerable skill, however, and should be performed only by the most skillful radiologists who are well versed in anatomy and procedures.

As one would expect, performing procedures in the mediastinum is contingent on comprehensive and complete knowledge of the mediastinal anatomy. One must be familiar with virtually all variations so that inadvertent puncture of a vascular structure does not occur. Aberrant subclavian vessels, left-sided vena cava, and prominent azygos veins are the most likely variations to occur. Appropriate use of bolus dynamic scanning is crucial to ensure that all vascular structures are well defined. Although it is seldom that tumors are vascular and represent a bleeding risk, it is common for vascular anomalies, aneurysms, or pseudoaneurysms to exist. Any rational interventionalist would use bolus dynamic scanning in the mediastinum to visualize normal anatomy and exclude a vascular anomaly ( Fig. 67-73 ); a more complete discussion of variant anatomy is provided in Chapter 38 .

Procedures of the different portions of the mediastinum (superior, anterior, middle, or posterior) require uniquely different approaches. Successful and safe procedures in these areas require careful attention to the subtle techniques used to guide and manipulate instruments in these areas.

Superior mediastinum.

The superior mediastinum consists of the space above the aortic arch and contains the various brachiocephalic vessels (venous and arterial). This is the most common area for congenital variations because arteries emanating from the arch and venous variations are common.

This space requires an anterior insertion in virtually all cases and usually results in an approach through the suprasternal notch or the parasternal region ( Fig. 67-74 ). In all cases a good vascular study with bolus injection is crucial to clearly define the carotids, jugulars, and small vessels. In most cases the lower portion of the thyroid gland and the trachea must be dealt with to approach such masses. It is always crucial to clearly identify the lobes of the thyroid because it is common for substernal or low-lying thyroid to be mistaken for a mass in the upper mediastinum. If the ipsilateral lobe of the gland is not identified, one should always consider the possibility that the observed mass might be the lobe of the thyroid.

Methods for choosing the correct entrance site and pathway are essentially the same as in other areas. As the needle is inserted, the pathway and tip of the needle are monitored carefully with incremental adjustments. Use of the bevel to slide past the different structures is important to provide clearance around the vessels. As noted in the earlier technical section, the bevel is turned toward the structure to be avoided, and the point is rotated so that it is directed toward the intended target. Also as noted in earlier sections, successful maneuvering of the needle around a vessel or structure is best accomplished by pushing the instrument slowly, which facilitates the deflection of the needle.

Anterior mediastinum.

Lesions in the anterior mediastinum are in the anterior space behind the sternum or anterior to the aortic arch. In most cases, one can access these lesions by using an insertion site just lateral to either side of the sternum. One must be cautious not to sample the internal mammary artery or vein. With these lesions, one can use either an aspiration or cutting needle as in other portions of the mediastinum. Careful localization of the needle tip or cutting needle gap is crucial in such cases ( Figs. 67-75 and 67-76 ) because cutting of the major vessels would be catastrophic. As noted in the earlier section, placing a patient in a lateral or oblique position can shift the anterior junction line to make such procedures simpler. One can inject saline into the mediastinum to increase the potential pathway for needle insertion and to avoid traversing the lung (see Fig. 67-51 ). Gupta and colleagues have proposed using a transsternal approach with a coaxial system; these authors advocate using a coaxial cannula to penetrate the sternum and use a needle through the cannula. We have not found the need for this approach and have preferred to use the methods discussed here.

Middle mediastinum.

The middle mediastinum is a large space below the arch and encompasses the azygos node ( Fig. 67-77 ), peritracheal region ( Fig. 67-78 ), aorticopulmonary window ( Figs. 67-79 and 67-80 ), subcarinal region ( Fig. 67-81 ), and periesophageal region ( Fig. 67-82 ). The peritracheal regions and the aorticopulmonary windows are typically best approached anteriorly through the anterior junction line, avoiding the lung parenchyma and other structures. The internal mammary vessels should be identified before any procedure so that the entrance and pathway can be planned appropriately to avoid these structures.

The anterior approach may be performed if the patient has a wide anterior junction region, but in many cases it is helpful to place the patient in a lateral position, either on the right or on the left side (see Fig. 67-76 ). By doing this, the anatomy may shift so that a clear pathway through the anterior fat space may be possible, preventing any possible pneumothorax that could occur if the lung were traversed (see Fig. 67-79 ). In such cases, injection of saline or gas may be helpful to enlarge slightly the potential pathway (see Fig. 67-51 ). Widening of the pathway is typically small because these materials disperse along the long axis of tissues and the cross-sectional areas, so the change in the anatomy is usually minimal. Maneuvering the instruments past corresponding vessels is a crucial part of the procedure.

Paratracheal or azygos lymph nodes can be approached by traversing the anterior mediastinum and the space between the superior vena cava and the ascending aorta. The level of the entrance is below the origins of the brachiocephalic vessels and the subclavian veins. When using this approach, it is crucial to choose the correct entrance site and trajectory to ensure uneventful passage of the needle past the vessels. Traversing the space between the vena cava and aorta is simpler if the process has slightly widened the space. In cases in which the vessels have not been separated in this fashion, successful placement can be accomplished by using a coaxial system as follows. With the newly designed devices, blunt or rounded stylets are available so that the walls can be pushed slightly and penetration can be avoided. In other cases, one might choose to use a small-caliber bevel needle to slide between the two structures, and the coaxial cannula can be advanced over the aspiration needle (see Fig. 67-77 ).

The subcarinal area is best approached through a posterior paraspinal approach from either the right or left side, depending on the location of the mass. The left approach between the aorta and the vertebral body is usually the best approach because there is more space compared with the paraspinal region on the right side (see Figs. 67-81 and 67-82 ). As one reviews the figures, the approach seems almost unbelievable, but the truth is that we have consistently performed this approach for 15 years with no complications. It would be gratifying to believe we were the only ones capable of doing this, but the facts are that the approach and methodology are valid, and most experienced and motivated interventional radiologists can learn to perform this procedure.

Choice of the correct pathway and entrance site is crucial to the success of the procedure. The most suitable trajectory from which an entrance site must be chosen is a pathway that includes the margin of the transverse process/rib junction, the margin of the vertebral body, and the mass itself. The trajectory path should not include any vessel. A small amount of lung parenchyma in the path is acceptable but not preferable. Selection of any other pathway more medial than this would be unsuccessful because the contiguous overlap of the transverse process and the rib articulation provides an almost impenetrable “shield” that is painful and traps the needle so that no angulation or adjustment is possible.

When this technique is executed, ample use of local anesthetic is important not only in the skin but also at the site of the transverse process and vertebral body. Unless adequately anesthetized, these two sites are painful because of the periosteum.

The steps of the procedure are as follows. The site is prepared and the needle is advanced just to the margin of the transverse process. At this level, local anesthesia is infiltrated on the periosteum. The needle is rotated so that the bevel is turned against the body, and the needle is pushed slowly. The tip of the needle is located close to the margin of the vertebral body, and local anesthesia is injected. The needle is adjusted if needed so that the bevel of the needle is against the edge of the vertebral body and pushed slowly several millimeters. The bevel is rotated so that it is toward the aorta and pushed forward 4 to 5 mm to pass the aorta. If the hemiazygos vein or esophagus is close, the bevel also may be adjusted to pass by those structures. When positioned in the mass, the sample is acquired as appropriate with either the aspiration or the cutting needle.

The sequence of events in this procedure typically proceeds predictably, but the procedure can be aborted any time at the judgment of the operator. As one begins to learn this technique, each succeeding attempt provides additional confidence that enhances eventual success and comfort with the procedure. One almost insurmountable difficulty is the presence of a large osteophyte that may occur at the edge of the vertebral body. When a large spur is present, one might not even choose to attempt the procedure. If the procedure is being performed and an obstruction occurs at the edge of the vertebral body, minor adjustments of the needle can be made to determine if passage is possible. Persisting beyond several attempts is seldom beneficial.

Posterior mediastinum.

Sampling abnormalities in the posterior mediastinum is much easier than in either the anterior or middle mediastinum. The posterior mediastinum represents the space adjacent to the vertebral body and the pleural region. In most cases, there is clear access to the mass without crossing the lung parenchyma. In such circumstances, it is usually most expeditious to perform a quick sample with an aspiration needle and follow it with a cutting needle. Performing the aspiration first permits one to ascertain the effectiveness of the local anesthesia and the cooperation of the patient before using the cutting needle.

Pneumothorax.

The most frequent complication for chest procedures is pneumothorax, which can be treated easily with a small-caliber tube and a one-way valve. The decision to insert a chest tube depends on patient symptoms, patient attitude, attending service, and patient reliability. In ideal circumstances, the chest tube is inserted only if the patient is symptomatic. Even a 50% pneumothorax does not require treatment unless the patient is symptomatic. If there is any question of the patient's symptoms, reliability, or housing status, the chest tube should be inserted. If a patient is healthy, one can attempt to defer chest tube insertion by performing a simple aspiration of the air and observing for a recurrence of a pneumothorax over several hours.

Simple aspiration of air from a pneumothorax may prevent the necessity of chest tube insertion ( Fig. 67-83 ). Yamagami and associates reported the effectiveness of simple aspiration in two different groups of patients. In their report on CT-guided biopsies, 283 consecutive patients were treated with either simple aspiration or chest tubes. Of a group of 52 patients with pneumothoraces, manual aspiration of the air was performed immediately after the biopsy; 91% had total resolution of the pneumothorax. If the amount of air aspirated was more than 543 cc, a chest tube was required. In their second report, Yamagami's group used the technique on pneumothoraces after RFA. In 129 patients treated with RF, there was a pneumothorax rate of 29.5%. Fourteen of these cases were treated by manual aspiration, and 24 were simply observed.

We use this approach in some patients when the pneumothorax is small and the patient has few symptoms. In such a case, if the repeat chest x-ray shows no recurrence, we discharge the patient without a chest tube as long as the patient is reasonably healthy, lives close to a healthcare facility, and does not live alone. Any recurrence or progression of the pneumothorax from the first chest x-ray (after air aspiration) to the second would prompt us to insert a chest tube.

Chest tube insertion is required if a patient becomes short of breath or the follow-up imaging shows a progressively enlarging pneumothorax, or both. It is best for an inexperienced operator to insert the chest tube under fluoroscopy, but when experience has been gained, it is important to be able to insert the tube quickly at the bedside. The placement of the chest tube is simple and can be accomplished in the following fashion. One should use an entrance site in the anterior second or third intercostal space on the side of the pneumothorax. After preparation of the site with antiseptic solution, one should thoroughly anesthetize the site down to and including the pleura. One can estimate the depth required for the tube insertion by using the anesthetic needle to aspirate air at the site.

After administration of the local anesthetic, a nick in the skin is made with a scalpel to provide ample space for the tube. As one pushes the tube through the chest wall, a “give” is noted; this indicates passage through the pleura. The tube is kept perpendicular relative to the axial plane until it passes into the pleural space. It is then angled slightly cephalad so that the tube lies in the apical lateral region ( Fig. 67-84 ). This location is optimal because in this position it would evacuate “rising” air when the patient is either lying down or standing up.

After placement of the tube, one should always reconfirm the proper location of the tube even if the placement was optimal. This can be done by checking the tube position with oblique or lateral films. Another method is to place the end of the valve close to one's ear and listen for the escape of air through the valve. After the placement of the tube, patients usually experience some pain as the lung reexpands against the tube. A single small dose of a narcotic agent relieves the discomfort, and it seldom recurs.

The tube is securely attached to the chest wall by suture and tape to prevent displacement. A complete explanation should be provided to the patient and nursing staff to prevent inappropriate or premature closure of the tube.

If the lung is reinflated for 10 to 12 hours, the tube can be removed as follows. The tube should be clamped for several hours and a repeat film taken. If the lung remains expanded, the tube can be removed. If not, the valve should be reopened for 24 hours and the procedure should be repeated later. If severe primary lung disease is present, expansion could take 2 to 3 days.

Hemoptysis and Intrabronchial Bleeding.

Coughing up a small amount of blood is of no great concern. In earlier times when only fluoroscopy was available and approaches were limited to direct anterior or posterior pathways, there was a higher incidence of hemoptysis. With that method, it was the accepted convention that a patient could cough up one third cup of blood without consequence.

Seeing this amount of bleeding is very disconcerting. The best management is to place the patient on oxygen and calmly reassure him or her. Repeat scans might be appropriate to ensure bleeding is not progressing. If bleeding persists, a resuscitation should be called for intubation and other supportive measures.

Treatment for intrabronchial bleeding is temporizing until a formal resuscitation can be performed with the anesthesia service. If a patient has severe hemoptysis, he or she should be placed in the ipsilateral-side-down position. This position lets blood pool in the impaired lung and protects the uninvolved lung, which is elevated. The patient should be given intranasal oxygen so that oxygenation is enhanced. An emergency page for anesthesia is appropriate so an endotracheal tube can be placed into the functioning lung. Support of the patient can be via ventilating the unimpaired lung.

Intrabronchial Bleeding and Air Embolism.

Intrabronchial bleeding is a serious problem with potentially fatal consequences, so immediate action is required. The appropriate action for the radiologist is to place the patient on oxygen, prepare for cardiopulmonary resuscitation (CPR), and call an immediate resuscitation with anesthesia support. The anesthesia team should intubate the patient, placing the tube in the nonbleeding side to tamponade the bleeding in the affected side.

The complication of air embolism was uniformly fatal until more recently. Two cases reported by Lattin and colleagues and Ohashi and coworkers noted the recovery of patients who were treated with hyperbaric oxygen. Based on this limited experience, if air embolism occurs, one should administer 100% oxygen immediately and use hyperbaric oxygen if available.

Preprocedure History, Review of Images, Planning, and Informed Consent.

Patient history and diagnostic scans are reviewed and informed consent obtained as noted in the earlier general section. Review of patient history, including liver failure, bleeding issues, and current medications, is important. Laboratory studies should be assessed to evaluate for coagulopathy; if issues exist, they can be addressed as noted earlier under “Coagulopathic Treatment.” Abnormal liver function should alert the radiologist to be observant for any capsular varices or typical varices that might be subject to bleeding if inadvertently damaged. Laboratory values for CBC and coagulation factors should be suitable, as discussed earlier.

Most liver biopsies can be performed as an outpatient procedure, assuming inpatient facilities are available as needed. After studying a series of 1000 patients, Perrault and coworkers advocated that virtually all liver biopsies can be performed as outpatient procedures if there are no complicating factors. Appropriate consent should be obtained and documented.

Review of prior scans helps preplan the procedure relative to possible approaches and patient positions. It saves time to pre-position the patient if a position other than supine is to be used. Any enhanced CT scan may show vascularity and other anatomic features that are important.

PET scan can show metabolically active areas ( Fig. 67-87A and B ) or necrotic areas for target planning, but prior surgery or ablation may produce findings suggesting neoplasm. Metastatic lesions are typically positive, but hepatocellular carcinoma may have a high false-positive rate. In some unusual cases, tumors may alter their metabolism so as not to consume glucose and may be PET negative (see Fig. 67-87C through E ) even if the biopsy is positive; any rapidly growing mass should be biopsied (see Fig. 67-87F ).

Hepatocellular carcinoma is less amenable to evaluation by PET using FDG. Many authors report an accuracy of detection ranging from 50% to 75%, and others have reported that FDG uptake correlates well with poorly differentiated tumors and tumors with fibrosis. In postoperative patients with increasing α-fetoprotein (AFP), Chen and associates found that 71% could be detected with FDG PET. In addition to FDG tracers for PET, other authors have reported the use of acetate and choline derivatives with good results. Additional studies are indicated before the role of these new agents can be defined.

Imaging Preferences.

Choice of modality between US, MRI, or CT must be made by the individual physician, but CT generally is still considered the gold standard. Difficult anatomic location, poor visualization under US, or concern about increased vascularity is a clear indication for CT guidance.

With increasing expertise, modern US refinements, and increased demand on CT scanning time, routine mass biopsies most often should be performed with US guidance. Cases that are difficult to visualize and access because of location or large body size are well suited to CT. Lesions in precarious locations, such as adjacent to the portal vein, gallbladder, diaphragm, or other site, may be triaged for CT guidance. Biopsy of patients with abnormal coagulopathy should be performed with CT or combined with fluoroscopy so that a hemostatic method can be used should bleeding occur. The changes produced by bleeding and coil insertion make US less useful for performance of the hemostatic coaxial/coil/thrombin technique.

Procedure Steps.

During the procedure, vital signs, neurologic status, and pulse oximetry should be monitored. Analgesics and sedative medications can be administered as needed (see earlier under “ Patient Cooperation and Sedation ”).

Target and trajectory determination.

A diagnostic series of scans is taken in anticipation of the procedure. From the current scans, the mass referred for biopsy is reevaluated. If the lesion is stable or larger, the procedure should proceed. If the lesion is significantly smaller, one should consider deferring the procedure until a later date after follow-up scans are performed. Triage to MRI or US likely should be performed if the mass cannot be well visualized on CT.

One selects the tentative target for biopsy and the appropriate pathway for the trajectory. If the mass is large, one selects a portion of the mass that is not necrotic and attempts to target a site at the interface of the tumor and normal tissue. In many cases, especially with primary liver tumors, the transitional area may provide more diagnostic information to differentiate the various cirrhotic nodules associated with transition from dysplastic tissue to neoplasm ( Fig. 67-88 ).

The trajectory pathway should be planned as follows. If the lesion appears mobile, which is common close to the diaphragm, the patient can be positioned with the ipsilateral side down, which may lessen motion of the diaphragm. The pathway should provide a small cuff of normal tissue (which tamponades any possible bleeding and prevents tumor seeding in needle track) and yet minimize the pathway through the liver.

Capsular lesions should not be approached directly but have an angled approach, including a cuff of normal liver (see Fig. 67-88 ); this has been shown to prevent tumor seeding and reduce the probability of bleeding (see “ Complications and Factors Affecting Incidence ,” later). The diaphragm and pleural space should be avoided if possible; it is best never to have a pathway more posterior than the midaxillary line or to use patient position to better display the pleural space ( Fig. 67-89 ). Although some operators have used the transdiaphragmatic approach, we do not recommend it.

Bolus dynamic contrast scan over the chosen target site should be performed to visualize and avoid any significant-sized vessels ( Figs. 67-90 and 67-91 ) and exclude excessive vascularity, as noted in the contraindications section. Lesions such as hemangiomas ( Fig. 67-92 ) can be detected and the procedure terminated. Intensely vascular lesions may bleed even with FNA ( Fig. 67-93 ) and cause complications.

Postprocedure Care.

Before removing the patient from the table, we take a repeat scan to evaluate for postprocedure problems. If any hematoma is noted, the clinical team is contacted and a type and crossmatch of blood products may be ordered if there is a concern that the bleeding is significant.

The postprocedure care of these patients is simple and consists of observing them for abdominal discomfort and monitoring vital signs for possible bleeding. The patient is kept on bed rest for 4 hours. Vital signs are checked every 15 minutes for 1 hour, every 30 minutes for 1 hour, and every hour for the next 2 hours. After 4 hours, if no issues arise, the outpatient can be discharged home with appropriate instructions. Patients should be instructed to rest quietly at home for 1 day with no strenuous activity. If any pain or other symptoms develop, the patient should seek medical attention.

Factors Affecting Accuracy.

Many factors affect the diagnostic results from liver biopsies. In our own experience with FNA needles, we limit our needle passes to two in virtually all cases. The expected yield should be about 75% to 85%. These factors include (1) extent of disease in liver and size of lesions, (2) type and size of needle used, (3) proper technique, (4) number of needle passes, (5) experience of the interventionalist, and (6) choice of modality.

The extent of disease has a great impact on biopsy results. If mass lesions are large, numerous, and accessible, the chances of a positive biopsy are greater. Conn and Yesner studied this factor with a series of blind biopsies performed on cadaver livers at autopsy. They graded the degree of involvement of the liver by the number of lesions and found the following recovery accuracy: grade I, 1 to 3 lesions, had a 16% success rate; grade II, 4 to 25 lesions, had a 27% success rate; grade III, 25 to 50 lesions, had a 45% success rate; and grade IV, with greater than 50% involvement by weight, had a 86% success rate. This finding is important to consider whenever a new method, technique, or type of needle is first being reported in the literature.

Proper technique includes accurate placement of the needle's sampling portion within the lesion, application of vacuum if needed, use of the appropriate needle device, and proper recovery of the tissue sample (see earlier under “ General Procedure Steps ”). There are some variations of different aspects, and technique can and should be modified according to the preferences of the operator, but a consistent technique is crucial.

The type of needle affects outcome, with cutting needles providing greater diagnostic tissue than cytologic samples. All authors reporting the use of cutting needles reported improved diagnostic accuracy and more specific information based on sophisticated histopathologic evaluation or assays.

Automated cutting needles are better than manual types for most interventionalists. Before their introduction, cutting needles were seldom used for biopsies because only a few radiologists had the dexterity to use the standard manual devices, such as the Tru-Cut. In a survey, more than 90% of radiologists performing liver biopsies used the automated devices. Hopper and associates, in the evaluation of numerous automated devices, noted that the gold standard device with which all devices should be compared and the one as yet unsurpassed is the 14-gauge manual Tru-Cut device ; such manual devices have a long learning curve, so automated devices are preferred.

Meticulous placement of the needle is critical regardless of the guidance modality used. With a standard aspiration needle, the needle end must be in the lesion and not adjacent to it. With a side cutting needle, the tissue receptacle of the stylet should be positioned at the interface of the focal lesion, with only a small portion traversing the normal rim of tissue. Better quality pathologic interpretation is likely if an interface between normal tissue and neoplastic tissue is provided for review and staining. Also, areas of definite necrosis, more noticeable with contrast material, should be avoided.

Larger-caliber needles provide more ample tissue, enhancing pathologic interpretation. Larger samples permit recutting of samples so that different modern biomarkers can be assessed. Such biomarkers are crucial for modern oncologic treatment, which targets specific enzymes, peptides, or membrane markers. The effect of needle size on diagnostic success has been well documented for aspiration needles in in vitro and in vivo studies. Andriole and colleagues studied the yield in liver biopsy of various aspiration needles and found that in vitro there was a definite increase in the diagnostic material with the larger caliber needles. In a series of liver biopsies, Pagani reported the results of a small clinical group using 22-gauge and 18-gauge needles. Pagani found a significant improved diagnostic accuracy, 85% versus 98%, without any difference in the complication rate. In our own institution, the diagnostic yield and complication rates of various aspiration needles have been determined and compared with cutting needles; these rates were reported by Martino and associates and Ha and colleagues. With a single pass with aspiration needles and providing samples for slide smears and cell block material, the diagnostic yield for malignant disease with a 22-gauge needle was 60.9% compared with 77.6% with a 20-gauge needle. Pagani's data were based on the performance of multiple passes, and our data involved single passes, which explains their higher yields.

A laboratory study by Plecha and coworkers compared the amount of diagnostic tissue obtained with the amount of local bleeding with various calibers of Tru-Cut needles (see Tables 67-2 and 67-3 ). They found that per unit of tissue, there was slightly more bleeding with the larger 14-gauge needle than with the 20-gauge needle. The data showed further, however, that if one looks at the relative tissue yield (reflected by the quantity of DNA) compared with the blood loss, there is relatively no difference among the various needle calibers (20 gauge, 18 gauge, and 14 gauge). Theoretically if one was required to obtain a finite amount of tissue, the efficiency and risk of using a 14-gauge needle may be less. As noted earlier, the amount of bleeding per unit of diagnostic tissue is no different among the needles. Also, significant bleeding is usually the result of inadvertently striking a small artery. This is essentially a random event, so that if multiple passes with a small-caliber needle are required, there is a statistically greater chance of bleeding. Three or four passes with a 20-gauge needle may be less safe than one pass of a 14-gauge needle.

It has not been clinically determined whether multiple small-caliber Tru-Cut needles are safer to use than a single pass with a large needle. In our practice, we have had a major complication with the use of a 20-gauge Tru-Cut, which resulted in a major hemorrhage and required a surgical procedure.

Multiple needle passes increases accuracy. From these data and the data of Ferrucci and colleagues, it is clear that the more needle passes taken, the higher the diagnostic accuracy. In their series, Ferrucci's group selected 20 cases that were positive and looked at the diagnostic yield for each needle pass. The yield was 75% with one pass and 90% with two passes. (One can never expect 100% yield even with five passes, because the data were selective; in the same article, Ferrucci's overall yield was 82%.) Even with blind biopsies, the diagnostic yield increases appreciably with multiple passes.

Experience of the operator directly affects the outcome of interventional procedures. Poon and colleagues reported on procedures of the liver, including biopsy and treatment, and found lower complication rate occurs with experience.