Our study highlights several barriers, in applying the revised Bethesda criteria to identify patients at risk for Lynch syndrome based on a diagnosis of an HNPCC tumor. These barriers include limited family histories captured by health care providers and inconsistent reporting of MSI-H histology for colorectal cancer specimens by pathologists. Patient compliance represents another barrier identified lack of insurance coverage for genetic counseling services.
The results of our study demonstrate that 38% of patients (145/380) have incomplete recording of their family history of cancer and/or poorly characterized histology, thus hindering the application of the revised Bethesda criteria to determine eligibility for MSI testing. For these 38% of patients, it remains unknown whether they would have met the revised Bethesda criteria, had more information been made available. While approximately 11% of all HNPCC-related tumors (41/380) diagnosed over the course of one year at our tertiary care hospital met revised Bethesda criteria, only 20% of these (8/41) were referred for cancer genetic counseling, despite an active Cancer Genetics Program within our institution. In some of these cases, the health care provider did discuss consideration of cancer genetic counseling, but for a variety of reasons, the patients declined. Furthermore, among those patients referred, a significant number did not follow through with the counseling due to potential concerns for insurance and employment discrimination, financial barriers created by insurance companies, and possibly, a lack of appreciation for the long term benefits of surveillance and primary prevention in families with a hereditary predisposition to cancer.
The first major barrier we encountered in conducting this study was the cursory documentation of family history data in the medical chart by practitioners. Age at cancer diagnosis was rarely documented for relatives, cancer histories for 2nd and 3rd degree relatives were not recorded, and cancer types were not always specified. We accessed our institutional Tumor Registry database, which collects and maintains family history data on all patients first diagnosed and/or treated for a malignancy. However, the data collected relies solely on family history information previously recorded in the patient chart. The Tumor Registry does not directly interview patients. Furthermore, the Commission on Cancer and the State Department of Health does not mandate recording specific details regarding family history. We found that detailed pedigree information for extended relatives was not available. As a result, the application of the revised Bethesda criteria depended mostly on the affected patient's past medical history information, as opposed to family history information.
The ability to identify patients and their families at risk for HNPCC is strongly dependent on the clinician's recognition, understanding and application of the revised Bethesda criteria. These clinical criteria require the compilation of a detailed and thorough family history, the simplest and potentially most essential diagnostic task in assessing patients. Not surprisingly, previous studies have shown that even when a family history was suggestive of HNPCC, the significance of the data was either not appreciated by the clinician or the potential relevance of multiple HNPCC-related cancers in the same individual was overlooked [22, 23].
We also noted that the Tumor Registry does not keep track of patients with non-melanoma skin cancers or precancerous lesions. We identified a total of 28 keratoacanthoma and sebaceous adenoma patients from the pathology database. In all instances dermatologists did not record a family history that could be found in the medical records. We identified one patient with a keratoacanthoma who met criteria #2 of the revised Bethesda guidelines since she was diagnosed with a metachronous colorectal cancer. This history, however, was not recognized by the dermatologist and the patient was not referred for cancer genetic counseling.
The low referral rate for cancer genetic counseling may, therefore, be partly attributed to poor recognition by clinicians of patients at-risk for HNPCC. Unlike several of the polyposis syndromes, HNPCC lacks striking phenotypic characteristics that would raise suspicion. Clinicians, therefore, must rely heavily upon a well-elucidated and well-documented family history, coupled with knowledge regarding the natural history of HNPCC and its association with multiple extracolonic tumors . In one cohort, endometrial cancer predated the diagnosis of CRC in over 50% of HNPCC female patients . Moreover, in the United States several cases have been successfully litigated when physicians failed to identify a patient at risk for a hereditary cancer predisposition syndrome .
Some practitioners may not appreciate the potential benefits of genetic testing for patients with cancer and their families. From a clinical management perspective, a number of studies have highlighted the importance of identifying patients and families with HNPCC. For example, in one study of patients with HNPCC, Järvinen et al.  showed that colorectal cancer surveillance reduced the risk and improved the survival of colorectal cancer in comparison to the control group who declined surveillance. Schmeler et al.  showed that prophylactic hysterectomy with bilateral salpingo-oophorectomy was an effective strategy for preventing endometrial and ovarian cancer in women with HNPCC. In comparison, Lindor and colleagues  demonstrated that patients who fulfill Amsterdam-1 criteria but without mismatch repair deficiency, and their kindred, can receive less intensive cancer surveillance than recommended for HNPCC patients, because these families have a significantly lower risk of colorectal cancer and no appreciable risk of extracolonic cancers, similar to the general population.
The second major barrier illuminated by our study was the inconsistent reporting by pathologists of specific features that define MSI-H histology repair. As a result of these histologic features not routinely being documented, it was not possible to analyze 20 of 111 colorectal cancer specimens diagnosed from 50 to 59 years of age (criteria #3). Following our request to retrospectively analyze these features for these 20 cases, the pathologist confirmed that 2 demonstrated tumor-infiltrating lymphocytes and 2 exhibited Crohn's-like lymphocytic reaction (20%). Multiple studies have demonstrated an improved prognosis in both sporadic and HNPCC cancers that demonstrate MSI-H histology [2, 27–33]. Further, mismatch repair status predicted the benefit of adjuvant fluorouracil chemotherapy in colorectal cancer [34, 35].
Pathologists play a central role in identifying HNPCC patients' age <50 years and HNPCC histology and could then reflex to MSI testing. This was proposed by Gologan et al.  in their prospective. Selecting specifically for age and histologic features, they found that out of 75 CRC cases examined, 23% were identified as MSI-H tumors. Furthermore, although the positive predictive value of histology suggestive of MSI-H was low at 32%, a total of 80% of the MSI-H CRCs had at least one of the criteria suggestive of MSI-H status . An electronic reminder system for pathologists increased MSH tumor and Lynch syndrome patient identification .
Lastly, among those patients ultimately referred to the Cancer Genetics Program, few followed through with a consultation. This represents the third major barrier for the identification of HNPCC individuals. Unfortunately, Medicare and other major insurers do not currently provide payment for the board-certified medical professionals who are specifically trained to provide genetic counseling about the risks, benefits and limitations of testing. The American College of Medical Genetics and the National Society of Genetic Counselors are actively pursuing appropriate recognition of non-physician genetics providers. So far, this has resulted in masters and doctoral-trained genetics providers being eligible to apply for a National Provider Identifier (NPI) through the Centers for Medicare and Medicaid in 2006 and the development of a CPT code (96040) for genetic counseling services, effective January 1, 2007. Patients may also have concerns regarding the potential for insurance and/or employment discrimination.
To circumvent the barrier identified by our study and others, a more universal approach to molecular testing for HNPCC has been advocated. To circumvent the lack of detailed family history to identify HNPCC patients, a model for MSI testing of selected CRC tumors was found to be efficacious and cost-effective, identifying 2.2 times more patients compared to current practice . In the study by Schofield et al. , MSI was used as an initial LS screening test followed by immunohistochemistry of MMR genes from CRC patients less than 60 years of age. From a cohort of 1,344 patients, 3.6% had germline mutations. Hampel et al.  evaluated the feasibility of screening for Lynch syndrome for all colorectal cancers. Limiting tumor analysis to patients who fulfill the revised Bethesda criteria failed to identify 28% of cases with Lynch syndrome. Lastly, the Evaluation of Genomic Applications in Practice and Prevention (EGAPP) Working Group commissioned an evidence-based review through the Agency for Healthcare Research and Quality to better understand the utility of DNA testing strategies in reducing morbidity and mortality from Lynch syndrome. Following a review of 104 studies published through 2006, they found sufficient evidence to recommend genetic testing for Lynch syndrome to all individuals with newly diagnosed colorectal cancer . This group concluded that such a strategy could lead to improvement in medical management and clinical outcomes not only for patients, but also their relatives with Lynch syndrome. The EGAPP Working Group recommendations should not be viewed as discounting the value of family history but instead, adding a new pathway to identify high risk families . The cost-effectiveness of four genetic testing strategies to indentify Lynch syndrome patients was developed from the U.S. healthcare system prospective. Universal testing would detect nearly twice as many cases as targeting younger patients with cost-effectiveness comparable with other preventive services .
Potential pitfalls to universal testing include the inability to definitively diagnose LS patients in an ever-changing genotype and phenotype landscape. For example, a recent series of LS families with MSH2 deficient tumors without detectable MSH2 mutations were found to have germ-line mutations in an upstream gene encoding for Ep-CAM . Would these individuals previously been counseled as familial CRC type X ? Expansion of the LS phenotype to include prostate and bladder cancer has been documented, which increases the number of persons to be tested[44, 45]. The ability of patients and at-risk relatives to deal with the emotional and psychological burdens of the complexity and uncertainty of genetic testing is understudied . Finally, the fragmented U.S. healthcare system does not provide a central repository to carefully delineate the cancer risk for specific gene mutations or alert carriers of mutations of unknown significance should future research alter today's understanding of cancer risk.