|Home | About | Journals | Submit | Contact Us | Français|
Pharmacogenetic testing can help identify patients with metastatic colorectal cancer more likely to respond to anti-EGFR therapy. We systematically reviewed the benefits and harms of EGFR-related pharmacogenetic testing of molecular targets downstream to KRAS in the treatment of metastatic colorectal cancer. We searched five electronic databases from January 2000 through November 2010, and conducted separate grey literature and conference abstracts searches. Two reviewers independently assessed all articles for relevance and quality. We identified 27 studies, primarily fair- to marginal-quality, small retrospective, and single-arm cohort studies with significant overlap in patient populations. We identified seven studies that studied BRAF in independent patient populations, one that studied NRAS, four that studied PIK3CA, eight that studied PTEN expression, and five that studied AKT expression. The best evidence for BRAF, NRAS, and PIK3CA comes from the largest retrospective study (n=649) of chemorefractory patients from seven European countries. In this study, BRAF mutation was present in 6.5% of KRAS wild-type tumors. Only 8.3% of persons with BRAF mutations, compared to 38% of persons without BRAF mutations (p=0.0012), responded to chemotherapy with cetuximab. Clinical sensitivity and the false positive fraction (1- specificity) were estimated at 9.8% (95% CI 6.3, 14.5) and 1.6% (95% CI 0.2, 5.6), respectively. BRAF mutation was also associated with worse median progression-free survival (absolute difference 18 weeks, p<0.0001), and overall survival (absolute difference 28 weeks, p<0.0001). In the only study comparing outcomes in persons who did (n=227) and did not (n=332) receive cetuximab with combination chemotherapy, those with BRAF mutation had worse survival outcomes regardless of whether or not they received cetuximab. Although NRAS and PIK3CA exon 20 mutations were also associated with worse outcomes compared to persons without these mutations, evidence is based on a small number of identified mutations. Evidence for protein expression of PTEN and AKT is more sparse and limited by variable methods for assessing protein expression. Low-quality evidence addressing clinical validity of pharmacogenetic testing in metastatic colorectal cancer patients suggests that BRAF mutations are associated with poorer treatment response and survival outcomes, although this association may be independent of treatment with EGFR inhibitors.
Colorectal cancer is the third leading cause of cancer-related death in the United States . Despite improvements in chemotherapy for metastatic colorectal cancer, overall five-year survival remains poor at just 11% for patients with metastatic disease . Currently, cetuximab (Erbitux®, ImClone Systems) and panitumumab (Vectibix®, Amgen) are approved by the U.S. Food and Drug Administration (FDA) for the treatment of metastatic colorectal cancer in the refractory disease setting [2,3]. These monoclonal antibodies bind to the EGFR, preventing binding and activation of the downstream signaling pathways, which are important for cancer cell proliferation, invasion, metastasis, and ne-ovascularization . Tumors with molecular alterations in the EGFR signaling pathway (e.g., mutations in KRAS, NRAS, BRAF, PIK3CA, loss of PTEN protein expression, or AKT over expression), however, may lead to a constitutively activated pathway not responsive to EGFR-targeted treatment.
Pharmacogenetic testing for KRAS has already entered clinical practice, such that persons with metastatic colorectal cancer whose tumors have KRAS mutations are not treated with EGFR monoclonal antibodies. In April 2009, ASCO issued a Provisional Clinical Opinion stating that all patients with metastatic colorectal cancer who are candidates for EGFR antibody therapy should have their tumor tested for KRAS mutations, and that persons with a KRAS mutation in codon 12 or 13 should not receive EGFR antibody as part of their treatment . In July 2009, the FDA revised the label for cetuximab and panitumumab to advise against use of these agents in persons with colorectal cancer positive for KRAS mutations . Even among patients with wild-type KRAS, however, the response rate to EGFR monoclonal antibodies is less than 20% . Primary research suggests that molecular alterations in the EGFR signaling pathway downstream to KRAS may also predict non-response to EGFR monoclonal antibodies. These alterations are less frequently occurring than KRAS (Figure 1), but testing for additional molecular alterations in those without KRAS mutations has the potential to identify other patients not likely to respond to anti-EGFR therapy before treatment begins, therefore preventing unnecessary treatment and associated harms and costs [8,7].
We systematically reviewed the evidence for the clinical benefit and harms of EGFR-related pharmacogenetic testing (downstream to KRAS) in predicting non-response to treatment with anti-EGFR therapy. We asked four key questions (KQ) (Figure 2):
KQ 1: In patients with mCRC, can other EGFR-related testing improve (or lead to non-inferior) patient outcomes or decision making compared to not using additional testing?
KQ 2: How well do each of these tests predict treatment effectiveness?
KQ 3: How well do each of these tests predict important health outcomes?
KQ 4: What are the potential harms to patients in using these tests to guide treatment decisions?
Studies were identified by searching electronic databases, conference abstracts, regulatory documents, and trial registries. MEDLINE was searched from January 2000 to November 2010 for English language abstracts. This search was adapted for four additional databases (Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Database of Abstracts of Reviews of Effects, and Health Technology Assessments Database) and limited to publications between 2000 and 2010, with no language restrictions. We also searched Conference Papers Index (via CSA) from 2009 to 2010 and hand searched selected scientific conferences from 2009 to November 2010. Relevant studies were also identified by searching clinicaltrials.gov, NIH RePORTER, Current Controlled Trials (International Standard Randomized Controlled Trial Number Register), WHO International Clinical Trials Registry Platform, and FDA regulatory documents via Drugs@FDA. The last search for all databases was performed on November 24, 2010. Search details are provided in Search Strategies document (Supplementary Table 1).
Two investigators independently reviewed 3,365 abstracts and 191 articles against a priori specified inclusion criteria (Figure 3). For all key questions, we considered studies that included persons with metastatic colorectal cancer being treated with cetuximab or panitumumab, either alone or in combination with other chemotherapeutic agents. Studies that only included patients with locally advanced disease were excluded. Testing included assays for mutations in BRAF, NRAS, PIK3CA, and protein expression for PTEN and AKT. We also included studies examining PTEN and AKT mutations and gene copy number. We excluded testing for EGFR protein expression or gene copy number; upstream molecular drivers (i.e., EGFR ligands epiregulin and amphiregulin); and molecular targets not directly part of the EGFR signaling cascade, but mediators in adjacent pathways. We considered any study reporting one or more of the following outcomes: overall survival, clinical response to treatment (e.g., progression free survival, time to progression), health-related quality of life, radiologic evidence of tumor progression, or potential adverse effects (e.g., incorrect genotype assignment leading to incorrect treatment assignment, delayed treatment, negative psychological effects, and ethical, legal, and social issues/concerns). We did not exclude studies based on study design or study quality. Excluded studies are listed in Supplementary Table 2.
Two investigators independently assessed the quality of each study using the quality criteria proposed by the EGAPP working group , supplemented by the Newcastle Ottawa Scale developed for observational studies , and reporting standards checklist (REMARK) developed for prognostic and predictive studies . Articles were rated good-, fair-, or marginal-quality. Good-quality studies were those that met the following criteria: prospective design; large, well-defined, and representative study population; genetic testing was described well; blinded assessment of genetic testing in relation to outcome; homogeneous treatment; low rate of missing data; sufficiently long follow-up; and well-described and well-conducted analysis of outcomes. Fair-quality studies did not meet all the criteria, but did not have any fatal flaws in study design. Marginal- or poor-quality studies had significant flaws or lack of reporting that implied bias affecting interpretation of study results. Disagreements about inclusion and quality were resolved by consensus with a third reviewer.
One investigator extracted all relevant data from the studies into evidence tables that included the following study details: critical features of study design and quality, funding source, patient characteristics (e.g., age, sex, race/ethnicity, Eastern Cooperative Oncology Group [ECOG] performance status, metastatic disease), treatment regimen and setting, genetic testing details (e.g., gene mutation(s)/protein expression, tumor sample, assay technique, scoring method), frequency of gene mutation or protein expression, a priori specified study outcomes (stratified by KRAS wild-type if available), and any outcome representing potential adverse effects. A second reviewer verified all extracted data.
We identified 27 studies (reported in 34 articles); however there was a significant overlap of populations studied. Our evidence synthesis focuses on studies with independent patient populations and, when possible, results in persons with metastatic colorectal cancer that have KRAS wild-type tumors. We focus on three primary outcomes— tumor response (or disease control if tumor response is not reported) based on radiographic findings, progression-free survival (PFS), and overall survival (OS). Studies either used RECIST or World Health Organization (WHO) criteria (based on radiographic findings) to assess tumor response or disease control. Most individual studies reported tumor response rates with or without an odds ratio (OR). For tumor response, we also calculated the true positive fraction (TPF or clinical sensitivity) and false positive fraction (FPF or 1-specificity) if sufficient data were reported in persons with KRAS wild-type tumors . For continuous outcomes (survival), in addition to reporting hazard ratios (HR), we also report absolute differences between groups in weeks or months of median progression-free or overall survival.
We summarize results qualitatively and provide these results in tables for easy comparison across studies representing unique populations. For tumor response based on imaging, we attempted quantitative synthesis (meta-analyses) for sensitivity, specificity, and odds ratios to evaluate the predictive value for each genetic test with sufficient data. Due to overlapping populations and lack of outcome reporting for individuals with KRAS wild-type tumors, only 3 studies could be included in the meta-analyses. We attempted bivariate analyses for sensitivity and specificity (of BRAF and PTEN testing) simultaneously , as well as univariate meta-analyses for sensitivity, specificity, and diagnostic odds ratios, separately using random effects models [14,15]. However, the small number of studies and clinical heterogeneity among studies prohibited us from producing meaningful combined estimates. We instead focused on the best available evidence (e.g., single large, well-reported study) to provide the best estimate of clinical validity. We also considered how additional studies with independent and overlapping populations confirmed, disagreed, and/or contributed additional information to the best evidence detailed.
In addition to a summary of evidence table, we also provide a summary table focusing on the strength of the body of evidence, based on the GRADE (Grading of Recommendations Assessment, Development and Evaluations) approach . The following four domains were assessed: risk of bias, consistency, directness, and precision. The overall strength of evidence was graded as high, moderate, low, or very low (insufficient).
We found no studies that directly assessed whether pharmacogenetic testing improves (or leads to non-inferior) important patient health outcomes (e.g., morbidity, mortality, health related quality of life) in metastatic colorectal cancer patients who received pharmacogenetic testing to guide EGFR monoclonal antibody treatment decisions, compared to those who did not receive pharmacogenetic testing (Key Question 1).
We found a total of 27 fair- to marginal-quality studies that evaluated pharmacogenetic testing for EGFR molecular targets downstream to KRAS and their association with tumor response (Key Question 2) or survival outcomes (Key Question 3) in patients with metastatic colorectal cancer treated with cetuximab or panitumumab [17-39]. Most trials had very limited reporting on important patient characteristics (Table 3) and were conducted in European countries. Patients in the few trials or centers that reported race/ethnicity were overwhelmingly white. Only two small studies included patients from the US [21,20], one of which included 23% non-white participants . One small study was conducted in South Korea . In studies that provided baseline patient characteristics, the age of patients ranged from 22 to 94 years, with the mean age ranging from 57 to 67 years. The cohorts of patients studied were 46 to 71% male.
Most studies evaluated response to cetuximab, either as monotherapy or in combination with other chemotherapy. Two studies included patients who received either cetuximab or panitumumab [27,24], and only one study included patients who exclusively received panitumumab . All studies, except for one, included a majority of patients who had received prior chemotherapy, or in some cases included exclusively patients identified as chemorefractory. This study by Tol and colleagues, a retrospective evaluation of the Dutch RCT CAIRO2, evaluated the addition of cetuximab to combination chemotherapy (capecitabine, oxaliplatin, bevacizumab) as first-line treatment in patients with metastatic colorectal cancer . Only eight studies reported the patients' performance score, and the majority of patients in these cohorts had no significant activity impairments (ECOG performance status of 0-1) [37,30,21,22,41, 28,40,18]. Among all patients studied, BRAF mutations ranged from 0 to 17%, NRAS 3%, PIK3CA 3 to 18%, loss of PTEN expression ranged from 12 to 42%, and loss of AKT expression ranged from 33 to 60%.
We found no good-quality studies. Most studies were retrospective single-arm evaluations of mutations or protein expression in cohorts of patients who received chemotherapy with cetuximab or panitumumab, comparing outcomes for persons with and without identified mutations or protein expression. Of the two studies that included both persons treated with and without cetuximab [18,20], only one reported outcomes comparing those receiving cetuximab versus those who did not . Most of the studies were small, with less than 100 patients included in the analysis. Therefore, these studies may not have been adequately powered to examine the predictive ability of less frequently occurring tumor mutations. Most of the studies evaluated cetuximab in various treatment settings, and many studies combined patients receiving different combinations of chemotherapy, with differing histories of prior chemotherapy. Some studies combined persons who received either cetuximab or panitumumab. All but one study were retrospective and stated a priori for inclusion that “sufficient” tumor sample had to be available. Some studies specified that tumor samples had to be EGFR expression positive. Details about how patients were selected were not reported, or were very minimal, in about one-third of the studies. In the single prospective study (n=110), only 73% (80 of 110) of the tumor samples were assessed for BRAF . No BRAF mutations were identified in this cohort. The level of reporting in the remaining studies was frequently inadequate to determine the proportion of missing data (from all persons eligible for the study). Information about unanalyzable or missing data (i.e., the number of and reasons for) was only reported in a few studies. Most studies did not provide sufficient data to determine if patients in the retrospectively-identified cohort were similar in terms of prognostic risk. Only a few of the studies performed analyses to determine if any important patient and treatment setting characteristics influenced the mutations' association with tumor response or survival. Studies rarely reported duration of study follow-up.
In addition to individual study quality concerns, included studies had significant overlap in populations studied. Of the 27 included studies, only seven studies evaluating BRAF (n=1,224) [17-22,40], one evaluating NRAS (n=649) , four evaluating PIK3CA (n=1,030) [17,18,20, 21], eight evaluating PTEN (n=742) [7,18,23, 25,26,28,29,31], and five evaluating AKT (n=249) [23-25,28,30] included independent patient populations (Figure 4). The largest study, conducted by DeRoock and colleagues, was a retrospective analysis of chemorefractory metastatic colorectal cancer patients who received cetuximab (n=649). This study provides the best evidence for the predictive value of BRAF, NRAS and PIK3CA . The best evidence for PTEN protein expression comes from a smaller retrospective study (n=173) . No studies of AKT expression reported outcomes stratified by KRAS wild-type.
Overall, we identified seven studies (n=1224) that reported tumor response outcomes [17-22,40], and three studies (n=968) [17,18,20] that reported survival outcomes, in independent patient populations. The best evidence for BRAF pharmacogenetic testing comes from one large, fair-quality retrospective analysis of a consortium of European patients with chemorefractory metastatic colorectal cancer who received cetuximab (Supplementary Tables 3 and and4)4) . This study by DeRoock and colleagues (n=649) included chemorefractory patients with sufficient primary tumor sample available from 11 centers in seven European countries who were treated with cetuximab in combination with other chemotherapy from 2001 to 2008. This cohort was mostly men (58%) with an average age of 61 years. Mutations from primary tumor samples were assessed using MassARRAY multiplex polymerase chain reaction (PCR), with a subset of samples independently validated using direct sequencing or allele-specific PCR. Ninety-five percent (350/370) of KRAS wild-type samples had BRAF mutation status assigned and outcome data available. BRAF mutation (V600E) was present in 6.5% (24/350) of KRAS wild-type. Only 8.3% of persons with BRAF mutations responded to chemotherapy with cetuximab (p= 0.0012), compared to 38% of persons with BRAF and KRAS wild-type tumors. Calculated true positive fraction (clinical sensitivity) and false positive fraction (1 -specificity) were estimated at 9.8% (95% CI 6.3 -14.5) and 1.6% (95% CI 0.2-5.6), respectively. BRAF mutation was also associated with worse progression-free survival (absolute difference 18 weeks, p< 0.0001), and overall survival (absolute difference 28 weeks, p< 0.0001). Odds ratios for tumor response (Supplementary Table 4) and hazard ratios for progression-free and overall survival (Supplementary Table 5) were adjusted for age, sex, previous chemotherapy, and treatment center. Findings from other studies with [42,30,7,34,35,32,37,24,26, 38, 29,41,20,39,18] and without [22,19,40, 20,18] overlapping populations were either not informative (because of low or no mutations identified, limitations in outcome reporting) or consistent with findings from the study by DeRoock and colleagues.
We identified only one retrospective evaluation of pharmacogenetic testing in an RCT of persons with metastatic colorectal cancer receiving chemotherapy with (n=227) or without cetuximab (n=332) as first-line treatment . In this study by Tol and colleagues, BRAF mutation was identified in 8.7% (45 of 518) of all persons. Outcomes based on imaging criteria were not reported in the subset of those with KRAS wild-type. For all patients, disease control was not statistically significantly different between persons with BRAF mutations versus BRAF wild-type, whether or not patients received cetuximab (Supplementary Table 4). For persons treated with cetuximab, median PFS was shorter for persons with BRAF mutation versus KRAS and BRAF wild-type, 6.5 months versus 11.4 months, respectively (absolute difference 4.9 months, p<0.0001). Results were similar for persons who received combination chemotherapy without cetuximab; median PFS was also shorter for persons with BRAF mutation versus wild-type, 5.7 months versus 10.8 months, respectively (absolute difference 5.1 months, p<0.0001). Overall survival, a secondary outcome in this study, showed a pattern consistent with that of PFS. Differences by BRAF mutation status for PFS and OS were essentially the same for both treatment groups (with or without cetuximab). PFS and OS were noticeably better for patients in this study (receiving first-line chemotherapy) compared with the chemorefractory patients in the two other retrospective analyses that reported survival outcomes, which suggests important clinical heterogeneity among studies [20,17].
Pharmacogenetic testing for mutations other than BRAF have less evidence. We found only one study for NRAS (n=649)  and four studies for PIK3CA (n=1030) [17,18,20,21]. Again, the best evidence comes from the largest multi-center retrospective analysis (n=649) by DeRoock and colleagues in chemorefractory metastatic colorectal cancer . In this study, only 82% (302/370) of KRAS wild-type tumors had NRAS status and outcomes. Four percent (13/302) of KRAS wild-type tumors had NRAS mutations and 13% (49/370) had PIK3CA mutations (exon 9 and 20). Although NRAS and PIK3CA exon 20 (not exon 9) mutations were associated with poorer outcomes, this evidence is based on a very small number of mutations from one study (Supplementary Tables 3 and and4).4). Overall, there was no statistically significant difference in PFS or OS between persons with tumors that had PIK3CA mutations versus wild-type (Supplementary Table 5). Authors conducted subgroup analyses (presumed a priori) for mutations in exon 9 versus mutations in exon 20 because of different proposed biological effects for domains encoded by these two exons. Compared with PIK3CA wild-type, PIK3CA exon 20 mutations, but not mutations in exon 9, appeared to predict poor tumor response and survival outcomes (Tables 3 and 4). The remaining two studies with independent patient populations did not report results by KRAS wild-type [20,21] (Supplementary Table 4). No other studies examining PIK3CA, with [18,20,33,7,37,38,24,29] or without  overlapping populations, report results for PIK3CA exon 9 and 20 separately.
Most studies focusing on molecular alterations in PTEN and AKT studied protein expression instead of mutations or gene copy number. Studies used immunohistochemistry (IHC) to examine protein expression of PTEN and AKT, but used different antibodies and scoring systems. We found eight studies for PTEN (n=742) [7,18,23,25,26,28,29,31], six of which (n=652) reported survival outcomes [7,18,23,26,28,29] (Supplementary Tables 3 and and4).4). There was some evidence to suggest that PTEN loss may be associated with non-response, though results are conflicting between studies. The best evidence comes from a retrospective cohort (n=173) by Laurent-Puig and colleagues . In this study, about 20% of the KRAS wild-type tumors had loss of PTEN protein expression. Loss of protein expression was not associated with tumor response or progression-free survival, but was associated with slightly worse overall survival (Supplementary Tables 3 and and4).4). Based on a small number of fair- to marginal-quality studies with differences in assay methodologies, PTEN expression does not appear to have clinically robust ability to predict survival response to cetuximab or panitumumab. Only five small studies (n=294) studied AKT protein expression [23-25,28,30], only two of which (n=194) reported survival outcomes [23,28] (Supplementary Tables 3 and and4)4) and none of which reported results for AKT loss in KRAS wild -type. None of the five studies showed a statistically significant association between AKT expression and tumor response or survival. Based on one study, PTEN and AKT protein expression are only concordant in 60% and 68% respectively, of primary and metastatic tumors .
We did not hypothesize any clinically significant harms to testing other than incorrect genotype assignment leading to incorrect treatment assignment (i.e., leading to subsequent withholding of potentially effective therapy, or giving therapy that has significant adverse effects and cost with little to no benefit). None of the included studies reported harms of testing, and we found no studies that explicitly addressed harms or that addressed psychological, ethical, legal, or social implications of testing.
The best evidence to estimate harms associated with incorrect treatment assignment based on testing for mutations in BRAF, NRAS, and PIK3CA comes from the largest retrospective study of chemorefractory patients with metastatic colorectal cancer by DeRoock and colleagues . Overall, the specificity of mutations in EGRF-related genes was very high, and therefore the false positive fraction (1-specificity) was low. These false positives are those few patients who would respond to treatment despite a mutation identified through genetic testing, from whom potentially effective treatment is withheld. Point estimates of false positive fractions for BRAF, NRAS, and PIK3CA exon 20 are 1.6%, 0.9%, and 0.0% of responders, respectively (Supplementary Table 4).
Of the studied molecular targets downstream from KRAS, the evidence is most promising for BRAF mutation as a negative predictor of response to EGFR monoclonal antibodies, and is most robust for persons with chemorefractory metastatic disease receiving cetuximab in combination chemotherapy (Supplementary Table 6). BRAF mutation is less common than KRAS mutations, approximately 5 to 20% versus 30 to 45%, respectively [43,44,45,46]. In the largest study, which was exclusively in chemorefractory patients, BRAF mutation was present in 6.5% of KRAS wild-type . The calculated true positive fraction (possible benefit), was 9.8% (95% CI 6.3-14.5), which meant that an additional (after KRAS testing) 9.8% of persons who did not respond to treatment were identified with BRAF testing. The calculated false positive fraction (possible harm) was 1.6% (95% CI 0.2-5.6), which meant that of those who responded to treatment, the proportion with BRAF mutation was small. BRAF mutation also was associated with worse median progression-free survival (absolute difference of 18 weeks) and overall survival (absolute difference of 28 weeks) in chemorefractory persons. One retrospective evaluation of an RCT that compared persons receiving cetuximab and combination chemotherapy to those receiving combination chemotherapy without cetuximab as first-line chemotherapy showed that persons with BRAF mutations had shorter progression-free and overall survival regardless of cetuximab, suggesting prognostic ability independent of treatment with cetuximab . While the overall magnitude of association (odds ratio) of BRAF mutation on tumor response and survival is similar to the association of KRAS mutation on tumor response and survival, the clinical sensitivity is much lower. In a recent good quality systematic review of KRAS testing in this clinical scenario, the sensitivity was 49% (95% CI 44-54) . In addition, the body of evidence for BRAF testing is much smaller and primarily comprises single-arm retrospective studies with poorly-characterized cohorts of patients (Supplementary Table 7). Similar to KRAS mutations, two studies that currently represent the best evidence for BRAF mutations showed association is greater in chemorefractory patients than in patients receiving cetuximab in combination with other chemotherapy as first-line therapy [17,18]. However, unlike KRAS, BRAF mutation appears to have prognostic ability independent of predicting response to cetuximab. Additional retrospective evaluations of RCTs comparing persons who received chemotherapy with and without EGFR monoclonal antibodies would help clarify the extent to which BRAF mutation predicts poor response to anti-EGFR therapy, or predicts poor prognosis independent of treatment effect. We identified conference abstracts, without full publication of results, of retrospective analyses of RCTs evaluating the addition of cetuximab to first-line therapy that also suggest that BRAF mutation in persons with metastatic colorectal cancer was a strong negative prognostic factor (independent of treatment effect) [47,48]. When these and other studies are fully reported, it would be important to attempt to clarify the prognostic significance of BRAF mutations in metastatic colorectal cancer, and separate out whether there is any additional pharma-cogenetic treatment selection role for BRAF mutation testing for anti-EGFR therapy in first-line versus second-line or higher treatment of metastatic colorectal cancer.
Important details that may the affect test accuracy and reproducibility of assays are not routinely reported in studies addressing clinical validity. The analytic validity of BRAF testing in colorectal cancer was not part of this systematic review. In general, we found that the published literature on analytic validity was sparse, and does not reflect the technology of the assays used in the studies. The analytic validity of BRAF testing in colorectal cancer is likely good, based on the one study by DeRoock and colleagues that independently validated assay results in a subset of patients using allele-specific PCR . Most applicable and least biased analytic validity evidence for BRAF testing should be available from proficiency testing programs, although the proficiency testing data would not address important pre-analytic factors (that relate to tumor specimen and dissection of tissue) that may also influence test performance.
Evidence for EGFR-related pharmacogenetic testing, other than KRAS testing, in metastatic colorectal cancer to guide the use of anti-EGFR chemotherapy is still very limited (Supplementary Table 8). Evidence for these tests comes almost exclusively from fair- to marginal-quality retrospective studies without a comparison cohort who did not receive EGFR monoclonal antibodies. In general, if the gene mutation is not uncommon, prospective studies are clearly preferred. If prospective evaluation studies are not available or feasible, then better retrospective studies are needed. These studies should be based on well described cohorts, either nested in trials or clinical settings with high quality of information (patient and outcome assessment and documentation), with good descriptions of patient and setting characteristics (in terms of prognostic factors and treatment), and good follow-up and measurement of patient outcomes.
Low GRADE retrospective observational evidence suggests that BRAF testing in metastatic colorectal cancer patients is a negative predictor of response and survival in those treated with cetuximab (Supplementary Table 8). However, it is unclear if the association of BRAF mutation with worse tumor response and survival is due to predicting response to treatment with cetuximab or prognosis independent of treatment. Evidence for NRAS and PIK3CA exon 20 is thus far based on a very limited number of tumors with identified mutations and needs to be replicated in other populations and treatment settings. Evidence for PTEN loss of expression is conflicting, and may be due to clinical heterogeneity or variation in analytic and pre-analytic factors. IHC assays (the antibodies and scoring system) for protein expression need to be validated and standardized. The evidence is rapidly evolving, with numerous relevant conference abstracts presented in 2009 and 2010, without full publication of results, which most certainly will add to the knowledge base about the clinical validity of these tests (Supplementary Table 9). Improved reporting of important patient characteristics, treatment setting, and details of assays and tumor sample will help inform the applicability and implementation of clinically valid pharmacogenetic tests into practice.
The authors thank Smyth Lai, MLS for conducting the literature searches, Rongwei (Rochelle) Fu, PhD for statistical consultation, and Kevin Lutz, MFA for his editorial support. They also thank the following reviewers for their contribution to this evidence review: B. Nedrow (Ned) Calonge, MD, MPH; W. David Dotson, PhD; Stephanie Melillo, MPH; C. Sue Richards, PhD, FACMG; Matthew Thompson, MD, MPH, D.PHIL; and Katrina Goddard, PhD. This review was conducted as part of the Comparative Effectiveness Research in Genetics in Colorectal Cancer, NCI RC 2-CA148471.
The authors declare that they have no competing interests.