Analysis of factors influencing molecular testing at diagnostic of colorectal cancer

The aim of the study was to evaluate the current rate of molecular testing prescription (KRAS codons 12/13, BRAF and microsatellite instability (MSI)) in newly diagnosed colorectal cancer (CRC) patients and to determine which factors influence testing. All incident CRC cases in 2010 were identified in the Poitou-Charentes General Cancer Registry. The exhaustive molecular testing performed was accessed in the French molecular genetics platform. Factors influencing prescription were analyzed using logistic regression. Among the 1269 CRCs included in the study, KRAS, BRAF and MSI testing accounted for 35.1%, 10.5% and 10.9%, respectively. KRAS testing was carried out in 65.5% of metastatic CRCs, and 26.1% of non-metastatic CRCs. Among metastatic CRCs, age (<60 years), site of primary tumour (left colon) and geographical area of treatment were factors related to KRAS testing. BRAF testing was contemporary to KRAS testing for 92.5% of patients. Factors related to MSI testing were age (<60 years), TNM stage (stage IV) and geographical area of treatment. Among CRC patients under 60 years old, only 37.5% had MSI testing. These results underscore the need to reduce disparities in CRC molecular testing and highlight the limited application of the French guidelines, especially concerning MSI testing.


Background
Colorectal cancer (CRC) is the third most common cancer worldwide [1]. To date, colorectal carcinogenesis has been classified in three distinct pathways: chromosomal instability (85%), microsatellite instability (MSI) (15%) and CpG island methylator phenotype (25%). MSI is related to a deficient DNA mismatch repair (dMMR) system due to germline mutation in a MMR gene in Lynch syndrome (LS), or more commonly to an epigenetic inactivation of MLH1 in sporadic cases. Approximately 45% of CRC cases have a KRAS mutation [2] and only patients with wild-type (WT) RAS metastatic CRC (mCRC) may benefit from anti-epidermal growth factor receptor monoclonal antibody therapy (anti-EGFR mAbs) [3,4]. A BRAF mutation (V600E) is present in approximately 12% of CRCs and confers a poor prognosis, especially in mCRCs [5][6][7][8]. In dMMR CRC, BRAF mutation is specific to a sporadic origin and eliminates a LS.
Since 2006, the French National Cancer Institute (INCa) has been supporting a national network of 28 hospital molecular genetics platforms throughout France, offering patients all essential molecular genetics techniques for all cancers. For CRC, KRAS (now complete RAS), BRAF and MSI testing are routinely performed. Since 2008, KRAS testing is supposed to be performed in all mCRC cases. Since KRAS and BRAF mutations are mutually exclusive [8], BRAF testing is performed only in KRAS WT tumours. In France, MSI testing is recommended in patients with a CRC at an age lower than 60 and/or if family history suggests a LS. Nevertheless, epidemiological data concerning these different testing procedures are lacking. A recent French retrospective study revealed that 81.1% of patients with a mCRC had KRAS testing [9]. This study has some limitations due the nonexhaustiveness of incident CRC cases included and patient recruitment based on physician willingness. The General Cancer Registry in the Poitou-Charentes region (GCRPC) covers an administrative region of 1.8 million people in south-western France (available at http://medphar.univpoitiers.fr/registre-cancers-poitou-charentes/) and has been collecting all incident cancer cases, thereby enabling exhaustive analysis of the molecular analyses (using INCa molecular cancer genetics platform) performed in all incident CRC cases. The aim of the study was to analyze routine practice of KRAS, BRAF and MSI molecular testing among all the CRC patients in Poitou-Charentes diagnosed in 2010.

Study population
Since 2008, the GCRPC has included all incident cases of cancer, involving subjects regularly residing in the Poitou-Charentes region at the time of diagnosis, whatever the place of care. The Poitou-Charentes region comprises four departments: Charente, Charente-Maritime, Deux-Sèvres and Vienne. The minimum items recorded in the GCRPC were demographic data, tumour characteristics and treatment. According to the French law the data collected from the GCRPC was approved by the CCTIRS (Comité Consultatif sur le Traitement de l'Information en matière de Recherche dans le domaine de la Santé, approval n°07-374) and the CNIL (Commission Nationale de l'Informatique et des Libertés, approval n°907,303). Using the GCRPC 1375 incident CRC patients were identified in 2010 and after exclusion of non-relevant cases, 1269 patients were included in the study (Fig. 1).

Molecular testing
In 2010, KRAS mutational status (exon 2 codons 12 and 13) was determined at the specific request of a clinician. Regarding BRAF mutational status (V600E), analysis was mostly performed by the INCa hospital molecular genetics platforms in case of KRAS wild-type status. MSI was to be determined at the specific request of the clinician (suspicion of LS) or by the platforms for patients under 60 years old. All of the exhaustive molecular analyses (n = 480) from the different hospital molecular genetics platforms were itemized (Poitiers (n = 401) and other platforms (n = 79)).

Statistical analysis
The aim of the study was to evaluate the rate of prescription of molecular testing (KRAS, BRAF and MSI) regarding guidelines applicable in 2010 to newly diagnosed CRC patients. Secondary objectives were to analyze which criteria influenced KRAS molecular testing for metastatic and non-metastatic CRC patients respectively, and which characteristics influenced MSI molecular testing for all CRC patients.
The study was conducted in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement. The descriptive statistics used for quantitative parameters were mean and standard deviation; for qualitative parameters were frequency and percentage. A logistic regression was carried out on factors that could promote KRAS testing and Fig. 1 Flowchart of the study. Abbreviations: CRC, colorectal cancer; WT, wild-type; MSI, microsatellite instability; MSS, microsatellite stable MSI testing and determined odds ratios (OR) with a 95% confidence interval (CI). The geographical area of primary treatment was defined from the location of the center where the first treatment of CRC was performed. Status of the center was categorized as public, private or university hospital.
Statistically significant factors derived from univariate analysis (P values <0.25) were selected for multivariate analysis using a stepwise descending selection procedure with a significance threshold at 0.05. Possible interactions between independent risk factors were tested by including proper cross-product terms in the regression models, and likelihood ratio tests comparing models with and without the interaction term were used to estimate the significance of the interaction. Data management and statistical analyses were performed using SAS software version 9.4 (SAS Institute, Cary, NC, USA).

Population
Between January 1st and December 31st 2010, 1269 incident cases of CRC were included in the study. The agestandardized incidence rates of CRC were respectively 38.3 per 100,000 in men and 26.9 per 100,000 in women. Mean age was 71.9 ± 11.8 years (Table 1). At diagnosis, 22.8% of CRCs were metastatic and 77.2% were nonmetastatic.

Molecular testing
Overall, 480 CRCs (37.8% of the cohort) had at least one molecular test (KRAS, BRAF or MSI). KRAS was mutated in 41.7% of cases (n = 175/420), BRAF mutation in 24.2% (n = 31/128) and a dMMR phenotype was found in 16.8% (n = 22/131). Among the 480 molecular tests in Poitou-Charentes incident cases of CRC, 83.5% (n = 401) Among overall cohort, age at diagnosis, site of primary tumor, stage at diagnosis, geographical area of primary treatment and status of the center were the factors related to KRAS testing (data not shown). Secondly, analyses of metastatic and non-metastatic CRCs were performed separately, given that KRAS testing is recommended only in cases of mCRC.
Among non-metastatic CRC patients, in multivariate analysis, age at diagnosis (<75 years; p < 0.0001), site of primary tumor (right colon; p = 0.026), stage at diagnosis (stage II and III; p < 0.0001), geographical area of primary treatment (p < 0.0001) and status of the center (private hospital; p < 0.0001) were factors related to KRAS molecular testing (Table 4). KRAS molecular testing for non-metastatic CRC patients was mainly requested by pathologists (n = 114, 44.7%) and surgeons (n = 72, 28.2%).

Discussion
Our study is the first one to simultaneously evaluate three molecular testing procedures (KRAS, BRAF and MSI) in CRC. Rates for these molecular testing procedures were systematically linked to age at CRC diagnosis, site of primary tumour, stage at diagnosis, geographical area of primary treatment and status of the center.
KRAS testing was performed in 35.1% of CRCs and as expected was more frequent in patients with a metastatic disease (65.5%). Although KRAS status is required for the anti-EGFR mAbs used in mCRC, there are few data on KRAS testing rates. In a French retrospective  [10]. Moreover, in our study when limited to mCRC patients receiving first-line chemotherapy, KRAS molecular testing rate was higher (83.9%). Among mCRC patients, in multivariate analysis young age at diagnosis, primary tumor located in left colon and geographical area of primary treatment were factors related to KRAS molecular testing. Frequent KRAS testing in young patients is probably explained by more "aggressive" treatment strategies in these patients, particularly anti-EGFR mAbs used. We have no explanation as to why KRAS testing was more frequent for left-sided tumors. KRAS testing was also significantly more frequent in the Vienne and Deux-Sèvres departments. In the Poitou-Charentes region there is only one university hospital located in the Vienne department. We can suppose that the higher rate of KRAS testing in Vienne department was linked to university hospital research programs and easier access to molecular testing. We observed that molecular testing procedures were mainly requested by pathologists and oncologists. An earlier request by gastroenterologists on initial biopsies should be encouraged to allow the availability of molecular tests results during the first oncological consultation in order to quickly define the optimal treatment for mCRC (RAS status and anti-EGFR treatment).
Our work showed that 26.1% of non-metastatic CRC cases had KRAS testing. The rate in the USA population is 5% [11]. Younger age, higher stage at diagnosis, geographical area of primary treatment and status of the center were factors related to KRAS molecular testing in non-metastatic CRCs. We can suppose that it was conducted at the request of the clinician to quickly begin appropriate treatment in the event of development of metachronous metastases, especially in stage III patients. In addition, for some pathologists it was easier to address pathological samples to a molecular cancer genetics platform at the time of the first pathological examination rather than later, when the tumor blocs were archived. To our knowledge there has been no previous study evaluating BRAF testing rates in CRC cases. In our study the rate of BRAF testing was 10.5% and the factors influencing BRAF testing are similar to those influencing KRAS testing. The rate of BRAF-mutated CRC (24.2%) was high as compared with the literature (approximately 12%) [12,13]. BRAF testing was mostly performed directly by molecular cancer genetic platform in patients with KRAS wild-type CRC since the two mutations are mutually exclusive. This point explains the high rate of BRAF-mutated CRC since only KRAS WT CRCs were analyzed for BRAF.
Concerning MSI testing, the rate seems low (10.8%) but the dMMR CRC rate is in accordance with literature data [14,15]. To our knowledge this is the first study that analyzing factors related to MSI testing rates. Like KRAS testing, MSI testing was associated in multivariate analysis with young age, higher tumor stage and geographical area of primary treatment. French guidelines suspicion of LS (dMMR CRC with no BRAF mutation) did not have an oncogenetic consultation (38.5%). We were not able to determine if this was due to patient refusal or if patients had not been addressed to an oncogenetic consultant by their referring physician.
The average time to obtain results of KRAS tests in our study was 30.6 days (between histological sampling and the date when the molecular test results were available in the platform). Lièvre et al., calculated the median delay between physician prescription and reception of the results as 23.6 ± 28.2 days, a delay somewhat shorter because measured differently [9]. In addition, in contrast to the Lièvre et al. study, our study is reflective of real The main strength of our study resides in the crossing of two reliable and exhaustive data banks, GCRPC and INCa molecular cancer genetics platforms. If none of the previous studies evaluating KRAS testing are as exhaustive, it is because they were based on incomplete database and/or on questionnaires sent to volunteer physicians. The main limitation of our work is the difficulty in extrapolating its results to other countries since CRC molecular tests are dependent on physicians' and pathologists' clinical practices. It is noteworthy that we accessed the molecular testing rates in 2010 since there is a delay of at least 2 years before obtaining high-quality CRC data from the GCRPC, a delay justified by the data collection process and the application of standards and requirements during case registration. Moreover, it is challenging to retrieve reliable and retrospective information on life-style and family history, but it would be interesting to complete this evaluation by including CCR risk factors which probably influences the choice of the clinician for ordering molecular testing. Finally, factors influencing these molecular testing procedures are relevant for countries which already performed these tests but also those who are implementing these tests in order to allow an optimal use, especially RAS testing for anti-EGFR therapy used in mCRC.

Conclusion
To conclude, this study is the first to provide a robust and exhaustive overview of molecular testing in CRC. As expected, we note a high level of KRAS testing in mCRC but also significant level in stage III CRC, which was probably undertaken in order to have KRAS results for patients with a high risk of disease recurrence. Moreover, MSI testing rate is low and not in accordance with French guidelines, which recommend systematic testing before the age of 60. In addition, these results highlighted on which factors it is possible to act to improve the molecular testing procedures essential to management of CRC patients, particularly MSI testing.