Limb Length Discrepancy

Plain Radiography

Teleoroentgenography

Teleoroentgenography is the simplest whole-leg radiographic technique for the assessment of leg length inequality. A radiograph of the entire lower extremity is obtained with the patient supine on the radiography table, with a long film and radiographic ruler beneath them. This technique is also the most susceptible to magnification error because a single exposure is made from a midpoint on the patient’s lower extremities ( Fig. 20.7 ). However, it is useful if the patient is unlikely or unwilling to hold still for multiple-exposure techniques.

Orthoroentgenography

Orthoroentgenography is a radiographic technique described by Green and colleagues in 1946. The purpose is to minimize measurement error resulting from magnification by making three separate exposures of the lower extremities centered over the hips, knees, and ankles ( Fig. 20.8 ). One long film with a radiographic ruler is obtained, similar to in the teleoroentgenogram. In patients with significant leg length inequality, separate exposures for each leg may be made, with the x-ray beam centered over each joint. It is important that the patient not move between exposures (Green and colleagues ensured this immobility by strapping the patient to the table after positioning) and that the limbs be aligned neutrally. Saleh and Milne described the use of this technique in a weight-bearing position so that angular deformities, mechanical axis deviations, and limb length can be assessed simultaneously in that position. Weight bearing is not important to the accurate measurement of leg length inequality.

Scanography

The term scanography arose from an early technique called slit scanography, but it has come to be used for a technique similar to orthoroentgenography. Scanography differs from orthoroentgenography in that, in addition to movement of the x-ray tube over the patient for three exposures, the film is moved under the patient ( Fig. 20.9 ). This approach reduces the size of the film required and makes storage and handling easier. As in orthoroentgenography, the patient must remain motionless during the exposures. The entire length of the bone is not available on film, thus negating this technique as a screening tool to assess the cause of leg length inequality. The EOS radiology technique offers accurate measurement of limb length discrepancy with minimal radiation.

Computed Radiography

An alternative to conventional radiography is low-dose computed radiography. The radiation dosage exposure with this method is significantly reduced (as little as 1% of the exposure with conventional radiography for orthoroentgenograms), and the accuracy is comparable to that of conventional radiography. ^,

Computed Tomography

CT has become popular for the assessment of leg length inequality. ⁱ

i References , , , , , , .

Advantages include lower radiation exposure even when the entire limb is exposed, greater accuracy, less susceptibility to error if the patient is poorly positioned, and the ability to accommodate positioning difficulties secondary to joint contractures or the presence of external fixators ( Fig. 20.10 ). This technique is specifically indicated when the patient has a knee flexion contracture or is in a circular external fixator because the accuracy is greater than that of conventional orthoroentgenography under these conditions. ^{, ,} The radiology technicians responsible for obtaining CT scanograms must position the patient’s limbs symmetrically or should at least ascertain that a given positioning difficulty will not interfere with accurate measurement. For example, hip or knee flexion contractures are well accommodated by CT when measurements of segment length are taken from the lateral projection, but only if the limb is in a neutral position with respect to abduction or adduction at the hip or there is no asymmetric frontal plane angular deformity.

Ultrasonography

Ultrasonography has been used for the assessment of leg length inequality. ^, Junk and colleagues found that this technique was more accurate than clinical assessment alone, and Terjesen and associates found the accuracy to be within 2 mm of standard radiographic techniques. However, special jigs must be constructed, and the patient must not move during the examination. The main advantage of this technique is that exposure to ionizing radiation is not required. We have no experience with this technique of assessing leg length inequality.

Alterations of Growth

In 1862, both Hueter and Volkmann noted that compressive forces in bone resulted in a slowing of growth and that tension increased bone growth and the formation of osseous tissue (the Hueter-Volkmann principle). ^, Wolff disputed this, believing that both compression and tension resulted in bone growth stimulation. Haas in 1945 and Gelbke in 1951 demonstrated that wire loops placed around the distal femur in dogs impeded longitudinal bone growth. Haas also noted that pins placed across the physis could restrict longitudinal growth. Gelbke stated that the controversy regarding the influences of compression and tension on longitudinal growth arose from a failure to distinguish between growing and mature bone and that bone tissue formation and bone growth were considered identical processes. When he placed loose and tight wire loops around the distal femurs of growing dogs, Gelbke found that longitudinal growth continued for several months, until pressure developed across the physis; the physis then narrowed radiographically and histologically, and growth inhibition followed. Gelbke also noted that these changes were reversible if the wires were cut after physeal narrowing and growth inhibition had occurred. He also attempted to produce traction on an apophyseal physis with wire loops, with less success. He concluded that the effect of tension on the physis does not increase enchondral bone growth and appears to have the same effect as compression.

Functional Physeal Anatomy

The physis is divided into horizontal zones termed germinal, proliferative, hypertrophic, and provisional calcification ( Fig. 20.11 ). The entire process of growth in this fashion—longitudinal growth by the cartilaginous “plate,” followed by ossification of the cartilaginous precursor—is known as enchondral ossification. Horizontal or peripheral growth of the physis occurs as well, in the specialized groove of Ranvier. The control of longitudinal growth and the mechanisms of cartilage cell hypertrophy, calcification, and ossification are still not thoroughly understood and are subject to many central and local hormonal and mechanical influences. The most widely recognized central hormonal regulator of physeal growth is growth hormone (GH). GH deficiency is associated with proportionate short stature, and excess GH secretion in the skeletally immature results in gigantism. GH action on physes is mediated by insulin-like growth factors produced in both the liver and physeal chondrocytes. Another important class of polypeptides (at least 10 classes have been identified) influencing normal physeal activity consists of the fibroblast growth factors, which have a mitogenic influence on physeal chondrocytes. A genetic defect in the fibroblast growth factor-3 receptor is responsible for achondroplasia.

Longitudinal growth of the epiphyses themselves is usually not taken into account when estimating the longitudinal growth of a particular long bone. Clinical and radiographic anthromorphometric measurements of the long bones during growth are usually made from the ends of the epiphysis; the contributions to longitudinal growth of the physes at both ends of the long bones of the upper and lower limb segments are known and are limited to the physes themselves. Thus, as Moseley has pointed out, growth data on the length of the leg and its femoral and tibial segments pertain to the entire length of the bone from epiphysis to epiphysis, but calculations regarding final length by virtue of growth from the physes ignore the increase in length provided by the epiphysis. ^{, , , ,}

In long bones with physes at each end, the contribution of each physis to the longitudinal growth of the bone is typically asymmetric (see Fig. 1.9). Digby studied the contribution of longitudinal growth by assessing growth arrest lines in anatomic specimens. Anderson and colleagues estimated the contributions to longitudinal growth of the distal femur and proximal tibia by assessing sharply delineated growth arrest lines on consecutive radiographs in a semilongitudinal study of 206 boys and girls. These investigators found that 71% of femoral growth occurred distally, and 57% of tibial growth occurred proximally.

The timing of the appearance of the secondary centers of ossification, which make the physis radiographically identifiable, also varies by location and, to some extent, by individual. Finally, the timing of closure with cessation of longitudinal growth in individual physes also varies by location and individual (see Figs. 1.5 and 1.7 for these approximations).

Anderson-Green-Messner Growth-Remaining Charts

Anderson, Green, and Messner published two articles that provide important information on the longitudinal growth of the femur and tibia and an estimation of growth remaining that can be used to time epiphysiodesis or stapling. ^, The first article, published in 1963, examined longitudinal growth data obtained in 100 children (50 boys and 50 girls) who were assessed at least once a year in the 8 years before their growth terminated. Fifty-one of the children were normal (25 girls and 26 boys), and 49 had poliomyelitis affecting only one lower extremity; the data from the normal legs of these 49 children were incorporated into the data from the normal children. Orthoroentgenograms (measuring the length of the femur and tibia, including each epiphysis), hand and wrist films for skeletal age (using Greulich and Pyle’s atlas ), total body size, and other variables were measured at each visit. Visits were scheduled close to the subjects’ birthdays, and measurements (including skeletal age) were recorded against chronologic age ( Table 20.1 ).

A number of interesting findings emerged from this study. The annual rate of overall growth (stature) rapidly decreased from birth to age 6 years and was stable from ages 6 through 9 years (average stature increment, 5.7 ± 0.93 cm). Femoral length increased at an average annual rate of 2.0 ± 0.27 cm, and tibial length increased at an average annual rate of 1.6 ± 0.23 cm. A pubertal growth spurt typically occurred sometime after age 9 years (and usually reached a maximum between 10 and 12 years of age in girls and between 12 and 14 years of age in boys). This growth spurt was followed by a final 4-year period of rapid decline in the rate of growth until cessation of growth ( Fig. 20.12 ). When average figures for stature and femoral or tibial length changes per year of chronologic growth were computed, the amplitude of the peak was blunted because the age at which the growth spurt occurred varied from one child to the next. In general, growth continued for 2 years after the adolescent growth spurt, irrespective of the age at which it occurred. As a consequence, although tables and graphs based on chronologic age are useful in younger children, because of this variation in the onset of the pubertal growth spurt these tables and graphs are not as valuable in children with 5 to 6 years of growth remaining.

Anderson and colleagues used skeletal age, as published by Greulich and Pyle, but noted that assessments from the elbow, hip, knee, and foot can also be used, as well as physical maturation parameters (such as Tanner’s stages of development; see Chapter 1 ). These authors found that when skeletal age was used as the basis for interval changes, the mean values for growth remaining were essentially the same as when chronologic age was used, but the recorded variation around the mean was appreciably less when skeletal age was used. Thus, more precise estimates of future growth in an individual child could be made by using skeletal rather than chronologic age, particularly in children whose level of maturation was consistently advanced or retarded 2 years or more compared with their chronologic age.

The method that Anderson and associates used to develop their growth-remaining charts was to interpolate the length of each of the 100 femora and tibiae at specific, equally spaced skeletal ages (10 years 3 months, 11 years 3 months, and so on). The increments of growth between these consecutive skeletal ages were derived. The growth of the entire femur and tibia was then related to the growth at the specific physes by using figures for the proportional contribution of growth obtained from the semilongitudinal study of sharply delineated growth arrest lines on consecutive radiographs (71% of femoral growth occurred distally, and 57% of tibial growth occurred proximally). These results were then used to compile the growth-remaining charts for boys and girls in the distal femur and proximal tibia ( Fig. 20.13 ). An example of the variations encountered was provided by Anderson and colleagues themselves, who noted that the 50th percentile growth remaining at the distal femur for a girl of skeletal age 10 years 3 months was 4.1 cm, whereas the extremes of growth that occurred at that age were actually from 2.2 to 7.2 cm; 80% of these girls were noted to grow between 3.3 and 5.0 cm.

Anderson and associates described how the growth-remaining charts were to be used in their 1963 article. Green and Anderson had presented less complete versions in 1947 and 1957. ^, The first consideration is the growth of the side not operated on. These investigators noted that the rate of growth inhibition in poliomyelitis is not constant over the entire period of growth and recommended determining the rate of growth in the 2 to 3 years before treatment by epiphysiodesis. In addition, if the physis of the affected leg is not growing at all, no correction should be expected, only prevention of further leg length inequality. The second consideration is the child’s relative maturity, the importance of which cannot be overemphasized. These investigators concluded that forming a total picture of each child is important, and they did not attempt to reduce the prediction of results to a precise, mathematical formula.

The second article by Anderson and colleagues, published in 1964, provided data on the length of the femur and tibia from a completely longitudinal study of 67 boys and 67 girls derived from radiographs taken on their birthdays from 1 to 18 years of age. The radiographs were teleoroentgenograms in younger children and orthoroentgenograms in older children. The authors used “appropriate” conversion factors on these 6-foot films to arrive at true femoral and tibial lengths. Recorded femoral length was “from the proximal articulating surface of the capital epiphysis to the most distal point on the lateral condyle,” and recorded tibial length was “from the mid-point of a line drawn across the proximal condyles to the mid-point of the distal articulating surface.” The data ( Tables 20.2 and 20.3 ) were used to generate length of femur and tibia graphs for boys and girls based on their chronologic ages ( Fig. 20.14 ). The difficulty of reconciling maturation and relative height can be seen in the authors’ example ( Fig. 20.15 ): longitudinal follow-up of a boy with hemihypertrophy showed a “dip” from 1 standard deviation above the norm to the norm during a period of presumed “delayed” maturation. Thus, assessment only during adolescence would give a false picture of the percentile length of the individual’s femur and tibia and lead to an underestimation of the amount of growth remaining during that time.

Menelaus Method

Menelaus reported on his experience with the White method (reported by White and Stubbins in 1944) in 1966 and reaffirmed this experience in 1981. ^, White had originally suggested that the distal femur grew ⅜ inch and the proximal tibia grew ¼ inch per year, with growth stopping at age 17 in boys and at age 16 in girls. Menelaus modified that assumption of growth cessation to age 16 for boys and age 14 for girls ( Fig. 20.16 ). He reported the results at skeletal maturity in 44 children who had undergone 53 epiphysiodeses using the Phemister technique and immobilization in plaster for 6 weeks. At skeletal maturity, 52% of the patients had leg length inequality within ¼ inch of the calculated discrepancy and 41% within ¾ inch of the calculated discrepancy; 7% had a residual discrepancy of more than ¾ inch. Menelaus stated that 89.6% of Green and Anderson’s patients were within ½ inch of the calculated discrepancy, compared with 80% of Menelaus’ patients.

Moseley Straight-Line Graph

Moseley in 1977 described a straight-line graph method for calculating the ultimate discrepancy in a skeletally immature child and determining the timing of long-leg epiphysiodesis to correct leg length inequality. His method is discussed further in other publications. The graph was constructed based on a mathematical reanalysis of the chronologic growth data on the length of the femur and tibia in normal boys and girls that were published by Anderson and colleagues in 1964. The purpose of Moseley’s graph was to simplify and improve the accuracy of calculations intended to estimate the ultimate discrepancy in growing children by incorporating into the calculations skeletal maturation based on hand-wrist bone films, growth inhibition, and relative size. In his words, “the growth of the legs can be represented by straight lines by a suitable manipulation of the scale of the abscissa.” As a consequence, with the nomogram of skeletal age to correct for percentile growth (i.e., relative size and skeletal maturation), the growth of the short leg is represented as a straight line, and the leg length inequality is represented as the vertical distance between the lines; the line indicating the growth of the shorter leg has a less steep slope compared with that of the longer leg, and the growth inhibition effected by epiphysiodesis can be indicated by altering growth inhibition (slope of growth of the longer leg) by the expected amount, based on the type of epiphysiodesis performed . The reference line in Moseley’s straight-line graph refers to the growth of the normal leg rather than the long leg (see Fig. 20.18 ). This is important because, in cases of overgrowth (e.g., hemihypertrophy), the abnormal leg should be plotted above the normal, shorter leg; this has a slight impact on the subsequent depiction of the limbs, the projected final discrepancy at maturity, and the timing of epiphysiodesis. (See Plate 20.1 on page 861 for a detailed discussion of the application of the straight-line graph method of Moseley.) In a review of 23 skeletally mature patients who had undergone epiphysiodeses for leg length inequality, Moseley found that his straight-line graph method yielded a mean error of 0.6 cm in predicting ultimate leg length inequality, compared with a mean error of 0.9 cm when the Anderson-Green-Messner growth-remaining method was used.

FLOAT NOT FOUND

Some assumptions are incorporated into the construction and use of Moseley’s straight-line graph. Skeletal age–determined lengths of the femur and tibia do not exist for skeletal ages younger than 8 years in the data of Anderson and associates. ^, Moseley assumed that individual patients maintained the same percentile skeletal age but acknowledged that this is not necessarily the case. In fact, Anderson and colleagues pointed out the tendency for longer femora to grow less and shorter femora to grow more with subsequent maturation (which is one reason that the accuracy of the Menelaus modification of the White-Stubbins method is comparable to that of the Moseley method). Finally, the straight-line determination of growth of the shorter (abnormal) leg presumes that growth inhibition is linear; however, as both Moseley and Shapiro pointed out, this is not necessarily true. Despite these problems, the accuracy of this method in practice is generally good, and we routinely use this method at our institution to determine leg length inequality at skeletal maturity and the appropriate timing of epiphysiodesis of the long leg.

Paley Multiplier Method

Paley and co-workers described the multiplier method to calculate leg length at skeletal maturity. ^{, ,} These authors identified an arithmetic factor (multiplier) by dividing femoral and tibial lengths at skeletal maturity by the femoral and tibial lengths at different ages during growth for each percentile group, by using a variety of available leg length databases. The length of each leg and the difference in leg length at maturity (assuming constant growth inhibition) can be calculated by multiplying these measurements by the appropriate multiplier for the subject’s age and gender ( Table 20.4 ). One analysis found that the multiplier method was not accurate compared to methods based on skeletal age when used for timing arrest in patients with vascular malformations.

Summary of Prediction Methods

The application of published longitudinal growth data and the methods of calculating growth remaining to determine the appropriate timing of epiphysiodesis for leg length inequality have several areas of potential inaccuracy. Identifying the cause of leg length inequality in children is important for determining the ultimate discrepancy at skeletal maturity. For example, total physeal destruction from infection, fracture, irradiation, or surgical ablation results in fairly predictable growth retardation from loss of growth of the affected physis for the duration of remaining skeletal growth. Congenital limb deficiencies characteristically cause consistent growth inhibition of the affected leg, so that the percentage of shortening remains fairly constant during skeletal growth even as the absolute amount of discrepancy increases. However, Shapiro pointed out that not all growth inhibition in growing children is linear (i.e., results in a constant percentage of growth inhibition or acceleration in the affected limb), as implied by the standard graphic methods of calculating ultimate leg length inequality. Eastwood and Cole noted a linear increase in discrepancy in only 8 of 20 patients treated by epiphysiodesis.

Shapiro described five basic patterns of leg length inequality development based on an assessment of discrepancy development in 803 patients ( Fig. 20.17 ). Type 1 is an upward slope pattern, implying a stable percentage of growth inhibition compared with the normal leg. Type 2 is characterized by an upward slope–deceleration pattern. Type 3 is characterized by an upward (or downward) slope–plateau pattern and is most typical of post–femoral shaft fracture overgrowth. In Shapiro’s series, growth in 85% of patients with overgrowth had plateaued by 3.5 years after fracture. Type 4 is an upward slope–plateau–upward slope pattern, seen only in abnormalities involving the proximal femur, such as septic arthritis, Legg-Perthes disease, and avascular necrosis (AVN) associated with the treatment of developmental dysplasia of the hip. Type 5 is characterized by an upward slope–plateau–downward slope pattern, meaning that an initially increasing discrepancy actually decreases with subsequent growth.

In Shapiro’s series, the type 1 growth pattern was typical of patients with proximal femoral focal deficiency, Ollier disease (enchondromatosis), congenital femoral deficiency with more than 6 cm of shortening, physeal obliteration from any cause, and poliomyelitis. Some patients with congenital femoral deficiency with less than 6 cm of shortening or with poliomyelitis tended to exhibit a type 2 pattern. Patients with idiopathic hemihypertrophy or hemiatrophy demonstrated type 1, 2, or 3 patterns. Only 31% of patients with overgrowth associated with vascular anomalies had type 1 growth, and the remainder had a type 2 or 3 pattern. Patients with neurofibromatosis most commonly had a type 1 pattern, but types 2, 3, and 5 were also seen. Similarly, patients with juvenile arthritis exhibited types 1, 2, 3, and 5 patterns. Patients with Legg–Perthes disease exhibited all five patterns.

Prediction of leg length inequality based on serial scanograms, hand and wrist radiographs for skeletal age, and straight-line graph analysis ( Figs. 20.18 and 20.19 ) has other potential pitfalls as well. ^j

j References , , , , , , , , , .

Kasser and Jenkins, in a study of normal leg growth in normal children between 5 and 10 years of age, found that, except in girls with advanced bone age, the use of Greulich and Pyle’s atlas in preference to chronologic age did not improve the accuracy of predicting leg length at skeletal maturity. There was a mean error of 2.4 cm by the Anderson-Green-Messner method and of 2.6 cm by the Moseley straight-line graph method when the prediction was based on skeletal age. Blair and colleagues found that only 22 of 67 patients treated by epiphysiodesis had a final discrepancy of less than 1 cm. Ten of the 45 failures resulted from inadequate epiphysiodesis, and 35 were caused by incorrect use of the growth prediction charts of Anderson and associates. Little and colleagues also found that use of Greulich and Pyle’s atlas and either the Anderson-Green-Messner charts or Moseley’s straight-line graph did not improve the accuracy of prediction using the Menelaus method alone, which is based on chronologic age.

Lampe and associates found that 9 of 30 patients treated by epiphysiodesis based on Moseley straight-line graph predictions had a discrepancy of more than 1.5 cm at skeletal maturity, including 1 patient operated on twice. These investigators concluded that the pattern of skeletal maturation limited the accuracy of the method. Beumer and colleagues, in a study of the growth of 182 Dutch children between 1979 and 1994, found that mean femoral and tibial length increased, compared with the data of Anderson and associates. Beumer and co-workers modified the Moseley straight-line graph based on this information and the use of skeletal age. Then, in a study of 34 patients treated by epiphysiodesis, these investigators found that the new graph better predicted limb length at maturity in 22 patients and yielded comparable results in 5 patients when predictions were made using the Moseley straight-line graph.

Carpenter and Lester evaluated skeletal age (according to Greulich and Pyle’s atlas) in the distal radius and ulna, carpus, and metacarpals and phalanges in 100 hand and wrist radiographs of 45 children younger than 10 years of age. These investigators found significant discrepancies between skeletal and chronologic age among the regions of the hand and between the sexes. Cundy and colleagues assessed variations in the designation of skeletal age in 60 radiographs by 4 radiologists who were using Greulich and Pyle’s atlas. These investigators found that 50% of children were assigned a skeletal age that differed by more than 1 year, and 10% varied by more than 2 years. As Ferron pointed out, the standard deviation of Greulich and Pyle’s atlas is “plus or minus one page.”

Epiphysiodesis