Molecular Testing of Gastrointestinal Neoplasms

Polymerase Chain Reaction–Based Microsatellite Instability Testing

Microsatellite instability testing uses surrogate loci to identify the mutator phenotype, or high-frequency MSI (MSI-H), in a highly sensitive manner. An unstable locus is defined as one in which the tumor DNA contains a new, or mutant, allele compared with the host germline DNA. MSI testing is accomplished by performing the MSI assay on normal tissue and a relatively pure sample of tumor DNA (generally from formalin-fixed, paraffin-embedded tissue with ideally >50% tumor cellularity) and comparing the electropherograms of allele sizes ( Fig. 13.1 ). The original “official” panel of microsatellites, known as the National Cancer Institute, or Bethesda, panel, recommended three dinucleotide repeats (D2S123, D5S346, and D17S250) and two mononucleotide repeats (BAT25 and BAT26); however, dinucleotide (e.g., CACACACACA…) repeats lead to a 5% to 10% rate of low-frequency MSI (MSI-L). MSI-L has been studied extensively, and currently, MSI-L CRCs have no different risk of Lynch syndrome and no different chemotherapy implications than microsatellite stable (MSS) tumors. A panel of mononucleotide repeats (e.g., poly-A) is preferred by most laboratories because it has virtually eliminated MSI-L phenotype and is likely more sensitive for detecting MSI-H, particularly those caused by MSH6 mutations.

Microsatellite Instability High Carcinoma: Sporadic versus Lynch Syndrome

Microsatellite instability high almost always occurs as a result of one of three processes—Lynch syndrome (and constitutional MMR deficiency), MLH1 promoter hypermethylation, or tumor-acquired somatic mutations in an MMR gene. Most commonly, an MSI-H carcinoma occurs because of epigenetic silencing of MLH1 promoter. MLH1 promoter hypermethylation is associated with the CpG island methylator phenotype (CIMP), the serrated pathway of colorectal carcinogenesis, and BRAF V600E mutations (∼75% of cases), not Lynch syndrome. Therefore, assessment of MLH1 promoter hypermethylation and/or BRAF mutation status is invaluable in helping determine whether a carcinoma with loss of MLH1 protein is sporadic or not. One must be aware, however, that MLH1 promoter hypermethylation can occur in Lynch syndrome as the “second hit” of MMR gene silencing, more rarely, in the setting of germline MLH1 methylation. BRAF V600E mutations are also reported in Lynch syndrome cancers but are about 1/10 as frequently as MLH1 methylation. Thus, BRAF V600E mutations are more specific for sporadic carcinomas with MLH1 loss but less sensitive than MLH1 promoter hypermethylation. Last, up to 50% of nonhypermethylated MSI-H carcinomas harbor tumor-acquired, rather than germline, MMR mutations. These “Lynch-like” or “tumor Lynch” carcinomas are difficult to identify. As few centers are testing tumors for somatic mutations in the MMR genes, Lynch-like carcinomas are usually a diagnosis of exclusion when no germline mutation is identified. However, a percentage of germline mutations are not detected with current methods. Therefore, these “Lynch-like” cases are almost certainly a mixture of undetected Lynch syndrome cancers and sporadic tumor-acquired mutations.

Mismatch Repair Immunohistochemistry

Most laboratories choose to screen for Lynch syndrome via IHC, mainly for logistical reasons: most have IHC capabilities, and most do not have PCR capabilities. Clinically, the tests are essentially equivalent such that an MSI-H by PCR is analogous to MMR protein loss of expression by IHC. In reality, the tests are complementary, and laboratories are best served when they can perform both assays because each has its strengths and weaknesses. In short, we prefer IHC when tissue is limited (e.g., biopsies, neoadjuvant therapy) and when rapid turnaround time is necessary. The immunostains are nuclear and essentially always have a built-in internal positive control. We prefer MSI testing on large, relatively pure samples of carcinoma (e.g., resections) because of its slightly better overall sensitivity and, counterintuitively, it is easier interpretation (see later). MSI-H results should be followed by IHC to determine the defected protein and help guide further testing.

The MMR stains may be difficult to interpret. Several studies have catalogued the relatively high observer variability in interpreting the DNA MMR immunostains. These studies usually conclude that some amount of experience is required to accurately interpret these stains or that all abnormal stains should be followed up by confirmatory MSI testing. In a perfect world, these IHC stains are “all or none,” with no shades of gray, but the interpretation may be challenging in some cases. Some pearls to prevent misinterpreting DNA MMR IHC are as follows:

• 1.

  Gain experience and some background knowledge of the stains. Experience clearly plays a role in interpreting these stains such that “experts” are more reliable than those without experience. These are nuclear stains; no other cytologic location of immunoreactivity counts as positive ( Figs. 13.2 and 13.3 ). Biopsies are easier to interpret than resection material, probably because of more consistent formalin fixation.

  
  

• 2.

  Understanding the protein pairings is quite helpful because the finding of one IHC stain may help confirm the result in its respective binding partner ( Table 13.3 ). MMR proteins function as heterodimers. PMS2 requires binding by MLH1 for its stability within the nucleus, and MSH6 requires MSH2 as its binding partner such that loss of MLH1 shows concomitant loss of PMS2, and loss of MSH2 shows loss of MSH6. However, “isolated” loss of PMS2 and MSH6 are valid results. Thus, intact IHC results for PMS2 and MSH6 alone confirm that all four MMR proteins are intact in nearly all cases while reducing testing costs. We often screen with only these two IHC stains on small biopsies when MSI results would be of uncertain reliability.

  Table 13.3

  Colorectal Cancer Biomarkers and Therapeutic Implications

  	Actionable Result	Frequency (%)	Implication
  MMR status	MSI-H or loss of IHC expression	15–20	Lack of benefit with 5-FU–based chemotherapy Benefit from PD-1 blockade
  KRAS/NRAS	Activating mutations in exon 2 (codons 12 and 13), exon 3 (codons 59 and 61), and exon 4 (codons 117 and 146)	50–60	Lack of benefit from anti-EGFR antibody therapy
  BRAF	V600E mutation	8–10	Lack of benefit from anti-EGFR antibody therapy
  CDX2	Loss of IHC expression	<5	Benefit from conventional chemotherapy

  EGFR , epidermal growth factor receptor; 5-FU , 5-Fluorouracil; IHC , immunohistochemistry; MSI-H , high-frequency microsatellite instability; PD-1 , programmed death 1.

• 3.

  Do not interpret the staining in tumor cells without adjacent positive internal control results. False interpretation of lost protein expression may occur if one examines areas without positive internal controls (e.g., stromal cells, lymphocytes, endothelial cells. A rare exception is constitutional MMR deficiency (discussed later).

• 4.

  Beware of overly aggressive antigen retrieval. Overstained slides may give the impression of falsely retained protein expression. In our hands and others’, this often manifests as cytoplasmic staining with a perinuclear “rim” of staining ( Fig. 13.4 ).

  

• 5.

  The stains may be patchy, especially MSH6 and after radiation therapy ( Fig. 13.5 ). Remember, the vast majority of true loss of expression occurs in every tumor cell.

  

• 6.

  Do not interpret IHC stains as “positive” or “negative” because these terms can lead to confusion or misinterpretation. “Intact” expression and “loss” of expression are more clear and accurate terms.

• 7.

  Lynch syndrome–associated carcinomas, rarely, if ever, arise from sessile serrated polyps/adenomas; therefore, testing of these lesions is extremely low yield when screening for Lynch syndrome.

Much rarer than Lynch syndrome is so-called constitutional MMR deficiency in which an individual inherits two mutations in an MMR gene rather than one mutation in conventional Lynch syndrome. These patients are characterized by early-onset malignancies, including hematologic and central nervous system tumors as well as Lynch syndrome–associated neoplasms. Because both alleles of a particular MMR gene are affected, immunostains for the particular gene product protein are likely to be lost in both the neoplasm and the background non-neoplastic tissue, and one always should consider constitutional MMR deficiency before deciding that the immunostain was technically unsatisfactory ( Fig. 13.6 ).

Microsatellite instability testing also has clinical relevance in helping oncologists predict chemoresponsiveness. Initially, oncologists used MMR status to help with stage II carcinomas in which there is only a marginal (~5%) benefit with adjuvant chemotherapy. For years, it has been known that MSI-H carcinomas do not respond to 5-fluorouracil (5-FU) chemotherapy, and oncologists often forgo 5-FU–based chemotherapy in “high-risk” stage II colon cancers (i.e., those with poor differentiation, angiolymphatic invasion, perineural invasion, invasion of serosa or another organ, perforation) that are MMR deficient. More recently, in a phase 2 study of patients with metastatic carcinoma, led by researchers at Johns Hopkins University, it was reported that clinical benefit of pembrolizumab, an anti–programmed death 1 (anti–PD-1) immune checkpoint inhibitor, was predicted by the tumor’s MMR status; the reported data supported the hypothesis that MMR deficient tumors are more responsive to PD-1 blockade than MMR proficient tumors. These results were followed-up in 2017 with a report of 86 patients with 12 different types of solid tumor malignancies that had progressed on conventional therapies, nearly half of whom had Lynch syndrome. Of these 86 patients, 46 (53%) had an objective response to pembrolizumab, including 18 (21%) with a complete response. These results, published in Science , led to the Food and Drug Administration (FDA) approval of pembrolizumab for MMR-deficient solid tumors (not just colorectal carcinoma) in May 2017. Type of testing—MSI versus IHC—is not specified by the FDA; therefore, the two methods should be viewed as equivalent, and principles in test selection and interpretation outlined in the preceding paragraphs still apply. Given that about 5% of solid tumors are MMR deficient, MMR testing is expected to play an expanded role in cancer care in the future. Last, preliminary data have shown that stage II CDX2-negative (by IHC) CRCs, although quite uncommon at about 5% of cases, may be another group of stage II carcinomas that would benefit from chemotherapy.

RAS , Extended RAS , and BRAF Testing

Anti-EGFR medications, cetuximab and panitumumab, are FDA approved for the treatment of patients with colorectal carcinoma with distant metastasis (i.e., stage IV). Although the original trials discovered that EGFR immunohistochemical detection did not predict response to therapy, the investigators quickly found that activating mutations in KRAS codons 12/13 (exon 2) were excellent predictors of nonresponse, prompting consensus recommendations that KRAS 12/13 should be tested and wild-type before initiating costly (~$80,000) and potentially toxic (skin rash) anti-EGFR therapies. Cetuximab and panitumumab work as competitive inhibitors of ligand binding and prevent receptor activation. It therefore stands to reason that activating mutations in RAS gene would not be affected by attempts at ligand-based inhibition. Interestingly, other EGFR inhibitors, aimed at the tyrosine kinase domain, are not effective against colorectal carcinoma, even those that are RAS wild type.

KRAS codon 12/13 mutations occur in about 40% of colorectal carcinomas. More recently, additional activating mutations in KRAS and NRAS have been shown to predict nonresponse to anti-EGFR monoclonal antibodies. These mutations, collectively known as extended RAS mutations, account for an additional 10% to 20% of patients. Current guidelines (American Society of Clinical Oncology, NCCN) call for extended RAS testing before anti-EGFR therapy with cetuximab or panitumumab. Codons 12, 13, 59, 61, 117, and 146 on both KRAS and NRAS genes are recommended on all stage IV tumors. Further, emerging data have suggested that patients with exon 2– or non–exon 2–activating mutations in KRAS or NRAS may be harmed by treatment with cetuximab or panitumumab, making timely extended RAS testing all the more important.

The utility of BRAF V600E mutation testing in terms of prediction of response to anti-EGFR therapy continues to evolve. BRAF -mutated tumors, particularly those that are MSS are difficult to study for two reasons: (1) BRAF mutations are uncommon in metastatic colorectal carcinoma, accounting for fewer than 10% of tumors, and (2) BRAF V600E mutation in combination with MSS phenotype engenders a significantly worse prognosis than BRAF wild-type tumors, so statistical survival benefit often translate to little or no clinical benefit. All that said, the pendulum appears to be swinging in the oncology community to use the identification of a BRAF V600E mutation in a similar manner as RAS testing (i.e., a BRAF mutation precludes anti-EGFR antibody therapy). For unknown reasons, BRAF inhibitor–based therapies are not effective against BRAF -mutated carcinomas.

Future Directions of Molecular Testing in Colorectal Carcinoma

Molecular oncology testing in colorectal carcinoma will become increasingly important in the future, particularly as cost per test decreases and throughput increases. Companion diagnostics (i.e., biomarker testing to predict response to chemotherapy) is expected to bear the brunt of the expanded testing focus. Next-generation sequencing (NGS), also known as deep sequencing or massive parallel sequencing, is the technology leading the charge. It allows for the relatively rapid and relatively low cost sequencing of multiple clinically relevant genes that may guide therapy or aid in diagnosis. For example, our institution performs NGS on all metastatic colorectal carcinomas to assess for KRAS and NRAS 12, 13, 61, 117, and 146 as well as BRAF codon 600 as a single multiplex assay rather than nine separate tests by Sanger sequencing, and this is just the tip of the iceberg. Similar technology is being applied to hereditary cancer syndrome diagnostics such as in patients with hereditary polyposis. A patient with a clinically determined polyposis can undergo testing for peripheral blood-based “polyposis” panel that searches for germline mutations in key polyposis genes such as APC, MUTYH, PTEN , SMAD4, BMPR1A , and STK11 .