Impact of exact segment by segment primary tumor location status on anti-EGFR antibody first-line treatment efficacy in RAS/BRAF wild-type and BRAF mutant metastatic colorectal cancer. A pooled analysis of AIO studies FIRE-1, CIOX, FIRE-3, XELAVIRI, and VOLFI.


 BackgroundPrimary tumor location (left vs. right) has prognostic and predictive impact on the therapeutic management of metastatic colorectal cancer (mCRC) in particular in the context of anti-epithelial growth factor receptor (anti-EGFR) antibodies. This analysis evaluates the relevance of exact segment-by-segment tumor location in patients with metastatic colorectal cancer on outcome and efficacy of anti-EGFR-antibodies.MethodsThis is a retrospective, pooled analysis of five randomized clinical trials (FIRE-1, CIOX, FIRE-3, XELAVIRI and VOLFI) treating metastatic colorectal cancer patients in a first-line setting, published between 2011-2019. Each trial was a multicentre, phase 2 or phase 3 trial in which patients with untreated metastatic colorectal cancer received chemotherapy regimens with or without monoclonal antibodies (anti-VEGF, anti-EGFR). Eligible were patients with histologically confirmed metastatic colorectal cancer in good performance status who were at least 18 years old. Individual data of 1809 patients with available exact primary tumor location were included into this analysis. Prognostic and predictive effects of primary tumor location were evaluated in uni- and multivariate analyses using the Kaplan Meier method, log rank tests, Cox regressions and logistic regressionsResults Exact primary tumor location is an important determinant of overall survival (OS) in mCRC patients (P<0.001). Multivariate analysis of RAS/BRAF wild-type metastatic colorectal cancer indicate that efficacy of anti-EGFR agents in terms of OS increases continuously from primary tumors located in the caecum (HR 2.63), ascending colon (HR 1.24), right flexure/transverse colon (HR 0.99), left flexure/descending colon (HR 0.91) to the sigmoid (HR 0.71) and rectum (HR 0.58), demonstrating significant benefit in sigmoid and rectal metastatic colorectal cancer, as well as clear detriment in caecum mCRC. Patients with BRAF V600E mutant disease arising from left-sided segments of the colorectum benefitted from EGFR-antibody treatment survival: hazard ratio for death in left-sided tumors: 0.42 (95% CI 0.19-0.92).Conclusions Primary tumor location of metastatic colorectal cancer affects prognosis. Anti-EGFR efficacy increases continuously from proximal to distal segments of the colorectum in metastatic colorectal cancer patients with RAS/BRAF wild-type and BRAF mutant tumors. Therefore, patients with BRAF mutant tumors of the distal segments may benefit from first-line Anti-EGFR-based therapy.Trial registrationFIRE1 trial registration ID n/aCIOX trial registration ID NCT00254137FIRE3 trial registration ID NCT00433927XELAVIRI trial registration ID NCT01249638VOLFI trial registration ID NCT01328171


FIRE1
trial registration ID n/a CIOX trial registration ID NCT00254137 FIRE3 trial registration ID NCT00433927 XELAVIRI trial registration ID NCT01249638 VOLFI trial registration ID NCT01328171 Background Primary tumor location (PTL), usually de ned as left-vs. right with a cut-off at the splenic exure, is a prognostic and predictive biomarker in the treatment of metastatic colorectal cancer (mCRC) [1][2][3][4].
Patients presenting with right-sided mCRC have a dismal prognosis as compared to patients with leftsided mCRC. Additionally the PTL also appears to predict the e cacy of antibodies targeting the epidermal growth factor receptor (EGFR) with patients presenting with left-sided mCRC deriving a substantial bene t in overall survival (OS) with cetuximab or panitumumab therapy [1,2].
Numerous efforts have been made to nd a clear molecular correlate of clinical observations that have established primary tumor location as a tool for clinical decision-making. Although various molecular factors occurring with differing frequencies throughout the colon and rectum can be described [5][6][7][8][9], a usable classi er still has to be established. A key problem in the development of a molecular-based "rightvs-left mCRC" classi er might be that clear cut-offs cannot be found for most molecular markers and the biology of molecular differences throughout the colorectum underlies rather continuous changes than dichotomous distributions [6,10].
Therefore, this pooled analysis of ve randomized trials evaluates the impact of the exact location of the primary tumor (i.e. caecum, ascending colon, right exure plus transverse colon, left exure plus descending colon, sigmoid and rectum) in patients with mCRC breaking up the dichotomy of left-vs. right-sided colorectal cancer into six subgroups, acknowledging the continuum hypothesis [10]. We ask the question to which extent the exact location in uences the prognosis also taking into account molecular subgroups as de ned by (K)RAS and BRAF mutations. Moreover, the effect of anti-EGFR treatment according to exact PTL is explored in patients with RAS/BRAF wildtype and BRAF V600E mutant mCRC.
As this investigation is a retrospective and unplanned analysis our ndings should be interpreted as such.

Trials
We performed a retrospective analysis of ve trials addressing patients with previously untreated mCRC.

Exact primary tumor location
Locations of the primary were extracted from the respective study report forms and differentiated into caecum, ascending colon, right exure, transverse colon, left exure, descending colon, sigmoid and rectum. Due to small numbers, tumors of the right exure and transverse colon were analyzed as one group as well as tumors originating from the left exure and the descending colon. Patients with more than one primary in more than one of these segments were excluded from the analysis of exact primary tumor location. Patients with tumors of the recto-sigmoid region were analyzed as sigmoid carcinoma.

Treatment
Treatment procedures were described in detail in the previous publications and are summarized in Table  1. Details concerning dosages are displayed in supplementary table 1. Treatments were strati ed according to anti-EGFR antibody use vs. chemotherapy alone or bevacizumab-based therapy.
De nition of progression-free survival and overall survival Progression-free survival (PFS) was de ned as time from randomisation to rst progression of disease or death from any cause (whatever occurred rst). In addition, the XELAVIRI study de ned PFS as time from randomisation until switching to a new anticancer drug. Overall survival (OS) was de ned as time from randomisation to death from any cause. Objective response (OR) was evaluated according to classi cations of the WHO (FIRE-1), RECIST 1.0 (CIOX, FIRE-3) or RECIST 1.1 (XELAVIRI, VOLFI).

Statistical analysis
All statistical analyses were performed using SAS software (version 9.4; SAS Institute, Cary, NC) and SPSS version 25.0 software (IBM Corporation, Amonk, NY, USA). Survival was expressed as medians including 95% con dence intervals by Kaplan Meier method and compared using log-rank tests.
Univariate and multivariate Cox regression analyses were used to conduct subgroup analyses. Additionally, Cox regression analyses with maximum-likelihood estimation were used for interaction testing. The two-sided signi cance level was set to 0.05 and estimates are reported with 95% con dence intervals.

Patient and tumor characteristics
The study population consists of a total of 1809 patients with known exact primary tumor location, including 1333 patients with known molecular characteristics (of those: 717 patients with RAS/BRAF wildtype tumors, 514 patients with RAS mutant tumors, 102 patients with BRAF V600E mutant tumors). An overview of contributing studies and frequency of each location is shown in Supplementary Fig. 1. Baseline characteristics of patients and tumors are summarized in Table 2. Several characteristics are signi cantly associated with the exact PTL. Of note, the frequency of BRAF mutations ranged from 10-20% among the characterized tumor specimens in the right-sided locations, whereas only approximately 3% of sigmoid and rectal specimens harboured such a mutation. In contrast, the frequency of RAS mutations is also pronounced in right-sided tumors and appears to decrease from right to left, but peaking again in rectal tumors. A graphical overview of baseline and tumor characteristics of interest with signi cant differences according to exact PTL is provided as supplementary Fig. 2.

Prognostic impact of exact primary tumor location
In the analyzed population, the numerically longest PFS and OS are seen in patients with primary tumors located in the left exure, descending colon and sigmoid. Among right-sided primaries no substantial differences were detected, despite primary tumors of right exure/transverse colon mCRC trended to be associated with the numerically shortest survival. Kaplan Meier curves for all patients are indicated in Figure 1. Uni-and multivariate comparisons of exact primary tumor location with rectal primaries used as reference con rmed right exure/transverse colon mCRC as the subgroups with the worst and clearly impaired prognosis in both uni-and multivariate analysis of overall survival. This nding appears to be driven by the two subgroups of patients with mutant mCRC (RAS and BRAF V600E). Please refer to Suppl. Figure 4 for details.

Predictive impact of exact primary tumor location in patients with RAS/BRAF wildtype and BRAF mutant tumors
The predictive effect of exact primary tumor location for anti-EGFR was explored in uni-and multivariate analyses (Kaplan Meier estimates of overall survival in Figure 2; logistic and Cox regressions in Figure 3). ORR appeared to be in uenced by exact primary tumor location with left sided locations being associated with a higher chance to achieve objective response to treatment. This observation was evident in both RAS/BRAF wildtype mCRC as well as in BRAF mutant mCRC. Clear bene t with anti-EGFR antibodies in terms of ORR was notably seen in left-sided mCRC with BRAF mutation in uni-as well as multivariate analyses.
In terms of PFS, a detrimental effect of anti-EGFR antibodies in patients with caecal RAS/BRAF wildtype mCRC as well as in patients caecal with BRAF mutant mCRC was demonstrated, whereas no signi cant differences in all other segments were evident with rather neutral (trend towards bene t in BRAF mutant mCRC) effects evolving with more distal primary tumor location. However, OS in RAS/BRAF wildtype mCRC appeared continuously improved with anti-EGFR antibodies the further distal the primary tumor was located. Whereas in patients with caecal primaries the usage of anti-EGFR antibodies resulted in a signi cant detrimental effect, neutral e cacy was seen in the continuity of the colon towards the sigmoid. The only subgroups demonstrating a clear bene t with anti-EGFR antibody treatment are patients with primaries located in sigmoid and rectal. Rectal PTL achieving the greatest bene t (hazard ratio for death: 0.58 (95% CI 0.43-0.77)). A similar pattern is observed in patients with BRAF mutant mCRC. An exploratory analysis in left-and right-sided BRAF mutant mCRC concerning anti-EGFR effects on OS suggests a signi cant bene t in favor of therapy with anti-EGFR antibodies in patients with leftsided, BRAF mutant mCRC. This is con rmed by both uni-and multivariate analysis and further con rmed by a signi cant interaction test (P for interaction=0.001). Please refer to Figures 2 and 3 for details. The interaction of PTL and anti-EGFR antibody e cacy is also present if the analysis is restricted to patients with BRAF mutant tumors from FIRE-3 and VOLFI representing the purest population as they contained direct randomization of anti-EGFR-antibodies (P for interaction: 0.003). Kaplan Meier curves of the FIRE-3/VOLFI subset are shown as Suppl. Figure 5. No trends towards favourable outcomes with the use of anti-EGFR antibodies were observed in patients with RAS mutant mCRC (data not shown).

Discussion
The presented analysis based on ve randomized trials including data from 1809 individual patients represents a large and robust basis to evaluate the prognostic and predictive effects of the exact primary tumor location in mCRC.
Whereas the usual differentiation between left-vs. right colorectal cancer is a known predictor of OS in mCRC [1,4,22], our analysis suggests that certain differences between the exact segments may exist. Of interest, a clearly dismal prognosis (as compared to primary tumors of the rectum) was detected in patients with mCRC deriving from primaries of the right exure/transverse colon. Similar trends of borderline signi cance in multivariate analyses were observed in patients with primaries in the caecum or ascending colon. Amongst patients with left-sided tumors, differences in outcome were not signi cantly different, although patients with rectal primaries appeared to achieve a numerically shorter OS as compared to patients with primaries of left exure/descending colon and sigmoid. The small advantage of sigmoid vs. rectal carcinoma might be in uenced by a smaller proportion of patients with RAS mutant mCRC in patients with sigmoid carcinoma (28% of tested specimens) as compared to patients with rectal cancer (42% of tested specimens) in context of a low frequency of BRAF V600E mutant tumors with dismal prognosis [23,24] in both segments.
The effect of anti-EGFR antibodies did not impact on PFS in our cohort, illustrating -in accordance with previous publications -that PFS does not necessarily re ect the e cacy of these drugs [19,25,26] and anti-EGFR antibodies may rather impact response related endpoints and OS [15,19,[27][28][29].
Clear effects of exact primary tumor location on overall survival were observed in patients with RAS/BRAF wildtype tumors. Interestingly, patients with primary tumors of the caecum derived a substantial disadvantage with anti-EGFR therapy, whereas substantial bene t was seen in patients with rectal cancers and -to a lesser extent -in patients with sigmoid primaries. Taken together, the data suggests a continuous increase of bene t from anti-EGFR antibodies from proximal (right) to distal (left) rather than the currently established dichotomous perspective of left (bene t from anti-EGFR) vs. right (no bene t from anti-EGFR) colorectal cancer [1,2]. This perspective is supported by molecular analysis of colorectal tumors that rather proposes a continuum of biological changes from proximal to distal segments of colon and rectum [6,10]. Based on these ndings, treatment of rectal and sigmoid RAS/BRAF wildtype mCRC with anti-EGFR-antibodies appears a clinical necessity. Particularly, mCRC patients with primaries in the rectum may achieve an enormous advantage with anti-EGFR-targeted therapy, and patients with mCRC with primaries in the sigmoid achieve a clinically relevant advantage.
Interestingly, in patients with BRAF mutant mCRC a similar pattern of increasing anti-EGFR antibody therapy e cacy from proximal to distal was found in the segment by segment analysis. Whereas rightsided tumors achieved no bene t, left-sided tumors appeared to bene t from anti-EGFR therapy in our population. This nding was con rmed by a multivariate analysis and a signi cant interaction test.
The idea that left-sided BRAF mutant tumors may respond to anti-EGFR treatment is further supported by an increased ORR and prolonged OS in the absence of clinically relevant effects on PFS in this data set. This combination has been frequently observed in an evidently anti-EGFR sensitive population (i.e. RAS/BRAF wildtype mCRC) [1][2][3]19].
Therefore, it could be hypothesized that BRAF mutation may not predict lack of anti-EGFR antibody e cacy. In turn, our ndings suggest that patients with BRAF mutant mCRC arising from left-sided primary tumors bene t to a similar extent from anti-EGFR antibodies as compared to RAS/BRAF wildtype tumors and should consecutively receive these drugs as part of rst-line therapy. It needs to be noted that the sample size of this pooled analysis is too limited to draw de nite conclusions on the bene t according to exact PTL (i.e. left exure/descending colon, sigmoid and rectum).
Our ndings are in contrast to reports that identi ed classical BRAF V600E mutations as likely negative predictors of anti-EGFR directed therapy [30][31][32]. However, it could be argued that these publications did not adjust their analyses for PTL and are consecutively biased by the relative over-representation of BRAF mutations in right-sided colon segments [33], a circumstance that is also present in our population.
Consecutively, current guideline recommendations [34] do not recommend anti-EGFR-targeted antibodies as part of rst-line therapy and cetuximab is currently used rather as combination partner for a BRAF inhibitor in refractory patients [35]. Accordingly, our nding in this particular subgroup challenges the actual treatment algorithm and needs con rmation from further trials with direct anti-EGFR-antibody randomisation such as CRYSTAL, PRIME, OPUS, CALGB/SWOG 80405 and PEAK [23,[36][37][38].
Challenging the idea of dichotomous left-right classi cation in mCRC, our hypothesis of a continuous increase in anti-EGFR therapy related bene t from proximal to distal parts of the colorectum supports the idea that the underlying molecular equivalent is likely a non-dichotomous biomarker (combination), potentially including EGFR-ligands, HER-3 messenger RNA or miR-31-3p [39][40][41][42].
Our pooled analysis is limited by its retrospective and exploratory design as well as by the low patient numbers in segments of the colon with an overall low frequency of primary tumors. Furthermore, a pool of ve studies (of those two directly randomizing the anti-EGFR antibody) using various treatment regimens and therefore, containing heterogeneous populations may have invoked potential undetected biases.

Conclusions
In our analysis, the exact primary tumor location of mCRC impacts prognosis and is also associated with increasing anti-EGFR e cacy in a continuous pattern from proximal to distal segments of the colorectum in patients with RAS/BRAF wildtype tumors. Similarly, patients with BRAF mutant mCRC originating from distal segments of the colorectum may bene t from rst-line anti-EGFR-based therapy. The hospital of the university (LMU) was the legal sponsor of FIRE-1 (supported by P zer), CIOX (supported by Merck-Darmstadt; Roche and P zer), FIRE-3 (supported by Merck-Darmstadt and P zer) and XELAVIRI (supported by Roche). The AIO Studien gGmbH was the legal sponsor of VOLFI (supported by Amgen). The supporting parties had no role in the design, analyses or decision to submit the data for publication.

List Of Abbreviations
Authors' contributions: