Tumor-Based Screening for Hereditary Nonpolyposis Colorectal Cancer: Does Age-Based Selection Optimize Yield?

Genetic predisposition testing for patients with several common cancers involves a blood draw for DNA sequencing and related assays. But laboratory testing of colorectal cancer tissue itself for evidence of microsatellite instability (MSI) has become the cornerstone in the process of identifying patients and families with hereditary nonpolyposis colorectal cancer (HNPCC; Lynch syndrome).^1,2 In the article accompanying this commentary, Gudgeon et al³ address selection of patient cases most likely to be informative through such testing. They show that limiting MSI testing to tumors occurring in patients before age 50 years does improve efficiency of testing by limiting the proportion of uninformative cases in which patients would otherwise be subjected to rather expensive testing. However, they suggest that the apparent cost saving comes at the expense of missing the small number of instances occurring in older patients that would be identified through a more universal approach to testing—an approach advocated by the Evaluation of Genomic Applications in Practice and Prevention Working Group.^4,5 They also emphasize the fact that the proportion of patients with informative MSI who actually follow through with mutational testing will affect the number of HNPCC cases ultimately detected.

A number of additional factors bear on the issue of efficiency in testing tumors for MSI. Most tumor evaluations start at the time of surgical resection. However, most patients with resectable rectal cancer undergo preoperative chemoradiotherapy, such that at least 20% to 30% of the time, there is insufficient viable tumor tissue remaining at the time of surgery.⁶ This supports the argument for reliance on the endoscopic biopsy performed at the time of diagnosis. One strategy is for the endoscopist to simply take a few additional biopsies of tumor, supplemented by biopsy of normal mucosa to serve as an internal normal reference for polymerase chain reaction–based MSI testing. Preoperative evaluation for MSI carries the additional potential for making a provisional diagnosis of HNPCC. This may influence surgical decision making with respect to extent of resection and/or performance of prophylactic hysterectomy or oophorectomy, virtually eliminating gynecologic cancer risk.⁷ A more common problem with tumor-based testing involves the patient who has previously undergone resection. Tissue blocks may no longer be available. Commonly, if referral is made to a medical oncologist at another institution, retrieval of sufficient tumor and normal reference tissue may be difficult. At our own institution, tumor testing in so-called medical oncology patients has been much less prevalent than in so-called surgical patients.

When universal testing is performed, a high rate of MSI occurs because of epigenetic silencing of the MLH1 gene by methylation of its promoter.⁸ This requires assays for methylation and/or BRAF mutation, as demonstrated by Gudgeon et al³ and many others.^9,10 Oversight of this process calls for close cooperation among the diagnostic laboratory, genetic counselors, and clinicians. At our facility, all MSI testing, including methylation/BRAF, is reviewed by genetic counselors attached to our GI center and also periodically reviewed by center surgeons, oncologists, and gastroenterologists. With respect to the point raised by Gudgeon et al regarding the proportion of patients following through with mutation testing, it must be echoed that the logistics of service provision do matter. Placing genetic counselors in point-of-service clinics, such as surgery, rather than medical genetics clinics seems to improve uptake of mutation testing. Clinical decisions regarding germline mutation testing will have to be made in compelling cases, such as those involving young patients with positive family history despite normal MSI/immunohistochemistry (IHC). If offered in such cases, it should be with the caveat that normal MSI/IHC carries powerful negative predictive value.

If universal testing is being considered at a given institution, decisions need to be made regarding the use, if any, of MSI data in cases that are likely sporadic (hypermethylated MLH1). Controversy continues over responsiveness of stages II and III MSI-high tumors to fluorouracil-based therapies.^11,12 Although this has nothing to do with HNPCC detection, it may serve as an independent justification for MSI testing in older patients.

Certain performance characteristics enter the equation when contemplating tumor testing. Approximately 3% to 7% of the time, there is a discrepancy between polymerase chain reaction–based MSI testing and IHC for mismatch repair protein expression.¹³ Although many centers rely on IHC only, occasional cases of HNPCC have been identified in which protein expression seems to be retained despite the presence of MSI.¹³ Thus, for purposes of case finding and quality assurance, use of both assays may be desirable, notwithstanding reservations about cost effectiveness and efficiency of such otherwise redundant testing.

A recent study by Moreira et al¹⁴ combined data from several population studies in which universal testing of tumors for MSI was performed. Sensitivity and specificity for ultimate mutation detection were compared under the universal testing strategy and compared with several clinical selection tools, including Bethesda guidelines. As one might suspect, universal testing carried the greatest sensitivity for mutation detection, but it did so with greater financial cost and a higher rate of false-positive MSI tests. The authors carefully considered the pros and cons of the various approaches, but in the end, they leave to the reader the value judgment: Is optimal sensitivity (through universal testing) in identifying all possible HNPCC cases the greater good, or is a more selective approach better, even if it means missing some patients whose relatives would benefit from predictive testing?

The debate will continue over which reasonable goal is more important: improved efficiency in detecting HNPCC through use of age cutoff, or ideal sensitivity for detecting all HNPCC cases through universal testing. The host of ancillary issues in MSI testing for HNPCC, as discussed in this commentary, may have some small bearing on the outcome of the debate.

Author's Disclosures of Potential Conflicts of Interest

The author indicated no potential conflicts of interest.