Oncogene mutation profile predicts tumor regression and survival in locally advanced rectal cancer patients treated with preoperative chemoradiotherapy and radical surgery

Abstract

Tumor response to preoperative chemoradiotherapy and postoperative survival differs among patients with locally advanced rectal cancer. The objective was to find correlations of mutated oncogenes and clinical outcomes in locally advanced rectal cancer. A total of 70 patients with preoperative preoperative chemoradiotherapy followed by radical surgery at a single cancer center between 2006 and 2012 were enrolled. Pretreatment tumor biopsy samples were assayed for 238 mutation hotspots harboring 19 oncogenes by time-of-flight mass spectrometry and OncoCarta Array. Oncogene mutations were found in 48.6% of patients (34/70). KRAS was the most frequent driver mutation, found in 35.7% of patients (25/70), followed by PIK3CA (14.3%), NRAS (5.7%), FLT3 (2.9%), and BRAF (1.4%). Multiple gene mutations were observed in eight patients (11.4%). Tumors with KRAS mutations responded poorly to preoperative chemoradiotherapy (p = 0.044). Patients with oncogene mutations had worse 3-year disease-free survival than those without mutations (67.2% vs 94.2%, p = 0.010). Patients with KRAS or RAS mutations had lower 3-year disease-free survival (68% vs 88.3%, p = 0.016; 65.5% vs 92.3%, p = 0.004, respectively) and 3-year overall survival (88% vs 95.4%, p = 0.020; 89.7% vs 94.9%, p = 0.036, respectively) than those without KRAS or RAS mutations. Oncogene mutation status affected tumor response to treatment and long-term survival in locally advanced rectal cancer.

Introduction

Preoperative chemoradiotherapy (CRT) followed by total mesorectal excision (TME) is the standard treatment of patients with locally advanced rectal cancer (LARC).^1,2 However, tumor response to preoperative CRT varies among patients. Previous reports show that the majority of patients have a partial response to CRT and that some ultimately achieve a complete pathological response (pCR).³ However, some patients have a poor tumor response and long-term oncologic outcome, indicating a lack of benefit from preoperative CRT.^4,5 In addition, we previously observed that a certain proportion of patients also tend to experience toxicity and treatment-related complications, including leucopenia, thrombocytopenia, radiation proctitis, anastomotic leakage, and poor wound healing.^6,7 Studies to identify molecular biomarkers predictive of clinical outcome of preoperative CRT in LARC patients have failed to identify any that are reliably associated with tumor response and long-term survival.^8,9

Matrix-assisted laser desorption ionization–time of flight mass spectrometry (MassARRAY) has high sensitivity and accuracy as a tool for genetic analysis.¹⁰ Previous studies of the relationships of genetic mutations and clinical outcome of cancer treatment have combined MassARRAY with OncoCarta panel assay.^11,12 In this study, we used this combined assay of oncogene mutation profiles to find correlations with short-term response to preoperative CRT and survival outcome in LARC patients.

Methods

Patient selection

This retrospective study included 70 LARC patients with preoperative CRT followed by curative surgery at Sun Yat-sen University Cancer Center between July 2006 and June 2012. Eligible patients had (1) histologically confirmed rectal adenocarcinoma, (2) clinical stage of T3–4 or N+ disease, and (3) radical resection of the rectal tumor. Patients with (1) missing surgical specimens or inadequate tissue, that is, paraffin blocks with <60% tumor cells, (2) metastatic disease diagnosed before or during preoperative treatment, or (3) other primary malignancy except for basal cell carcinoma of the skin were excluded. Clinical characteristics, treatment response, and follow-up results were retrieved from patient medical records and follow-up tracking system. This study was conducted following the ethical standards of the World Medical Association Declaration of Helsinki and was approved by the Sun Yat-sen University Cancer Center Institutional Review Board. Before receiving treatment, patients gave informed consent to use their tissue samples.

Treatment

All patients underwent preoperative irradiation of 50 Gy, delivered over 5 weeks in fractions of 2.0 Gy daily on 5 consecutive days per week. Concurrent chemotherapy was administered as two cycles of the XELOX regimen (intravenous oxaliplatin 130 mg/m² on day 1 and oral capecitabine 1000 mg/m² twice daily on days 1–14 in each 3-week cycle). All patients underwent a standard radical resection 6–8 weeks after the completion of preoperative radiotherapy. Adjuvant chemotherapy including six cycles of XELOX was scheduled to begin within 3–6 weeks postoperatively.

Pathological evaluation

Tumor tissues were obtained from formalin-fixed, paraffin-embedded (FFPE) pretreatment biopsy specimens. Hematoxylin and eosin (HE)–stained sections were reviewed independently by two pathologists. Paraffin blocks with more than 60% tumor cells were chosen for DNA extraction. HE-stained sections of postoperative FFPE tissue specimens were assessed using the Ryan tumor regression grade (TRG) system.¹³ Briefly, TRG 0 or TRG 1 indicated a significant response; TRG 2 or TRG 3 indicated a poor response.

DNA extraction

DNA was extracted from eight 5-µm sections from each FFPE sample using a QIAamp DNA FFPE Tissue Kit (QIAGEN, Hilden, Germany) following the manufacturer’s protocol. The quality of the isolated DNA was verified using a Nanodrop ND-2000 Spectrophotometer (Thermo Scientific, Niederelbert, Germany). The total DNA extracted from each sample was diluted to a concentration of 10 ng/mL for further analysis.

OncoCarta assay

A set of predesigned and prevalidated assays (v 1.0; Sequenom Inc., San Diego, CA, USA) were used for identification of 238 candidate mutations harboring 19 cancer-associated genes: ABL1, AKT1, AKT2, BRAF, CDK, EGFR, ERBB2, FGFR1, FGFR3, FLT3, HRAS, JAK2, KIT, KRAS, MET, NRAS, PDGFRA, PIK3CA, and RET. The mutations of each oncogene are shown in Supplemental Table S1. Briefly, 24 different OncoCarta polymerase chain reaction (PCR) primer sets were used to amplify the DNA samples. Extension primers were added to perform extension reactions. The products were aliquoted onto 384-well SpectroChipII using a MassARRAY Nano-dispenser RS1000 (Sequenom Inc.) and analyzed by matrix-assisted laser desorption/ionization–time of flight (MALDI-TOF) mass spectrometry (Sequenom Inc.). High-performance liquid chromatography (HPLC) with purified water was a blank control and normal human leukocytes were a negative control. The assay protocol is shown in Supplemental Table S2. Matrix chips were analyzed with MassARRAY Typer software (v4.0; Sequenom Inc.) with a predefined cutoff mutation frequency of 1%. Successful experiments were confirmed by detection of a standard peak and a blank control with no sample peak (Supplemental Figure S1).

Statistical analysis

Data were analyzed by Statistical Package for the Social Sciences (SPSS, version 17.0; IBM, Chicago, IL, USA). The significance of correlations between confirmed mutation status and tumor regression was assessed by the chi-square or Fisher’s exact test, as appropriate. Kaplan–Meier analysis and the log-rank test were used to compare differences in survival rate between subgroups with frequently mutated genes (≥10%). Overall survival (OS) was calculated from the date of surgical resection to the date of death or last follow-up. Disease-free survival (DFS) was the interval from surgical resection to tumor recurrence’ or death or last follow-up. Clinical follow-up was completed by March 2016. All tests were two-tailed; p < 0.05 was considered statistically significant.

Results

Patient characteristics

Of the 70 included patients, 47 were men (67.1%) and 23 were women (32.9%), with a median age of 55 (range: 28–75) years. The clinicopathological characteristics of the patients are summarized in Table 1. The median inferior tumor margin was 5 (range: 2–8) cm from the anal verge (DAV).

Table 1.

Clinical characteristics of locally advanced rectal cancer patients.

Characteristics	Number of patents (%)
Gender
Male	47 (67.1)
Female	23 (32.9)
Age (years)
Median (range)	55 (28–75)
<60	45 (64.3)
≥60	25 (35.7)
DAV (cm)
Median (range)	5 (2–8)
<6	43 (61.4)
6–10	27 (38.6)
Pre-CRT clinical stage
II	27 (38.6)
III	43 (61.4)
Pre-CRT clinical T stage
T2	2 (2.9)
T3	41 (58.6)
T4	27 (38.6)
Pre-CRT clinical N stage
0	27 (38.6)
1	23 (32.9)
2	20 (28.6)
Pre-CRT CEA level (ng/mL)
≤5	51 (73.9)
>5	18 (26.1)
Surgical procedure
Anterior resection	48 (68.6)
Abdominal perineal resection	21 (30.0)
Hartmann procedure	1 (1.4)
TRG
0 (complete response, pCR)	37 (52.9)
1 (moderate response)	10 (14.3)
2 (mild response)	10 (14.3)
3 (no response)	13 (18.6)
Evaluation of regression
Significant	47 (67.1)
Poor	23 (32.9)
Postoperative chemotherapy (weeks)
<12	19 (27.1)
≥12	51 (72.9)
Recurrence	17 (24.3)
Local recurrence	6 (8.6)
Liver metastasis	1 (1.4)
Lung metastasis	6 (8.6)
Other metastasis	5 (7.1)

CRT: chemoradiotherapy; DAV: inferior tumor margin from the anal verge; TRG: tumor regression grade; pCR: complete pathological response; CEA: carcinoembryonic antigen.

Mutation profiles

Genetic mutations were detected in 48.6% of the LARC tumors (34/70) and mutations of five oncogenes were detected (Figure 1). KRAS was the most frequent driver mutation, found in 35.7% of the patients (25/70), followed by PIK3CA (14.3%), NRAS (5.7%), FLT3 (2.9%), and BRAF (1.4%). Overall, RAS mutations, including KRAS, NRAS and HRAS, occurred in 41.4% of the patients (29/70). No mutations were found in the remaining oncogenes except for one activating mutation in exon 3 (Q61L). All KRAS mutations occurred in exon 2. The G13D amino acid substitution was the most common KRAS mutation (14.3%, 10/70 patients), followed by G12D (10.0%, 7/70 patients). The mutation loci of the detected oncogenes are summarized in Supplemental Table S3. Eight patients (11.4%) had multiple mutations, and the details are shown in Supplemental Table S4. Both KRAS and PIK3CA mutations were observed in 75.0% of these patients (6/8) and were the most frequent coexistent mutations.

Figure 1.

Frequency of multiple oncogene mutations.

[Figure omitted. See PDF]

Tumor regression and oncogene mutation

Tumor regression and oncogene mutation status are shown in Figure 2 and Table 2. In total, 34 of the 45 patients (75.6%) with wild-type KRAS achieved significant tumor regression (TRG 0 or TRG 1) after preoperative CRT, whereas only 13 of the 25 patients (52.0%) with mutant KRAS had a detectable response (p = 0.044). No significant correlations of tumor regression and other oncogene alterations (RAS, PIK3CA, NRAS, FLT3, and BRAF) were found. Multivariate logistic regression analysis indicated that KRAS mutation was independently associated with tumor regression after CRT (odds ratio = 0.332, 95% confidence interval: 0.112–0.982, p = 0.046; Table 3).

Figure 2.

Histologic picture showing rectal adenocarcinoma on biopsy specimen before preoperative chemoradiotherapy (CRT) and tumor regression grade (TRG) 0 and 3 after surgery by original magnification ×20 and ×100.

[Figure omitted. See PDF]

Table 2.

Comparison of tumor regression and oncologic survival outcome based on oncogene mutation status.

Gene mutation	Poor regressiona	Significant regressiona	p value	Overall survival			Disease-free survival
	n = 23 (%)	n = 47 (%)		Deaths (n)	3-year rate (%)	p value	Events (n)	3-year rate (%)	p value
Detected oncogeneb
Yes	14 (60.9)	20 (42.6)	0.150	9/34	91.0	0.197	13/34	67.2	0.010
No	9 (39.1)	27 (57.4)		4/36	94.2		4/36	94.2
RASc
Yes	13 (56.5)	16 (34.0)	0.073	9/29	89.7	0.036	12/29	65.5	0.004
No	10 (43.5)	31 (66.0)		4/41	94.9		5/41	92.3
KRAS
Yes	12 (52.2)	13 (27.7)	0.044	8/25	88.0	0.020	10/25	68.0	0.016
No	11 (47.8)	34 (72.3)		5/45	95.4		7/45	88.3
PIK3CA
Yes	2 (8.7)	8 (17.0)	0.480	2/10	77.8	0.870	3/10	68.6	0.632
No	21 (91.3)	39 (83.0)		11/60	94.9		14/60	82.8
NRASd
Yes	1 (4.3)	3 (6.4)	1.000
No	22 (95.7)	44 (93.6)
FLT3d
Yes	1 (4.3)	1 (2.1)	1.000
No	22 (95.7)	46 (97.9)
BRAFd
Yes	1 (4.3)	0	0.329
No	22 (95.7)	47 (100)

a

Significant regression was defined as TRG 0 or 1; poor regression was defined as TRG 2 or 3.

b

Detected oncogenes included KRAS, PIK3CA, NRAS, FLT3, and BRAF.

c

RAS included KRAS, NRAS, and HRAS.

d

The mutations of NRAS, FLT3, and BRAF were too rare to perform comparative survival analysis.

Table 3.

Univariate and multivariate analyses of indicators for tumor significant regression to preoperative chemoradiotherapy in locally advanced rectal cancer.

Variable	Univariate	p value	Multivariate	p value
	Odds ratio (95% CI)		Odds ratio (95% CI)
Gender (male)	0.623 (0.206–1.880)	0.401
Age > 60 years	0.803 (0.286–2.254)	0.667
T stage 4	0.415 (0.157–1.096)	0.076
N stage 1/2	0.679 (0.252–1.830)	0.445
CEA level > 5 ng/mL	0.378 (0.125–1.148)	0.086
CA199 level > 35 U/mL	0.650 (0.133–3.188)	0.596
DAV > 6 cm	0.326 (0.116–0.918)	0.034	0.310 (0.106–0.910)	0.033
KRAS mutation	0.350 (0.124–0.990)	0.048	0.332 (0.112–0.982)	0.046

CI: confidence interval; DAV: inferior tumor margin from the anal verge; CEA: carcinoembryonic antigen.

Oncogene mutation and survival

A total of 17 patients (24.3%) ultimately developed a postoperative recurrence during a median 50.3-month follow-up (range: 1–113 months). The 3-year DFS and OS rates were 80.8% and 92.6%, respectively. The survival of patients with different oncogene mutation status is shown in Table 2. Patients with any mutation in the detected oncogene had worse 3-year DFS compared with those without mutations (67.2% vs 94.2%, p = 0.010; Figure 3(a)), whereas the two groups had comparable 3-year OS (91.0% vs 94.2%, p = 0.197; Figure 3(b)). Patients with KRAS oncogene mutations had a lower 3-year DFS rate than those with wild-type KRAS (68% vs 88.3%, p = 0.016; Figure 4(a)) and a lower 3-year OS rate (88% vs 95.4%, p = 0.020; Figure 4(b)). Patients with mutated RAS also had worse 3-year DFS (65.5% vs 92.3%, p = 0.004) and 3-year OS (89.7% vs 94.9%, p = 0.036) rates than those with wild-type RAS. Additionally, patients with KRAS G12D mutations had worse 3-year OS rate than those without that specific locus mutation (71.4% vs 95.1%, p = 0.028).

Figure 3.

Kaplan–Meier curves of (a) 3-year disease-free survival (DFS) rate and (b) 3-year overall survival (OS) rate and oncogene mutation status in patients with locally advanced rectal cancer.

[Figure omitted. See PDF]

Figure 4.

Kaplan–Meier curves of (a) 3-year disease-free survival (DFS) rate and (b) 3-year overall survival (OS) rate and KRAS mutation status in patients with locally advanced rectal cancer.

[Figure omitted. See PDF]

Discussion

In this study, a profile of genetic mutations of 19 oncogenes primarily related to the RAS-RAF-mitogen-activated protein kinase (MAPK) and PI3K-AKT pathways revealed clinical and statistically significant correlations with outcome in LARC patients with preoperative CRT and curative surgery. KRAS (35.7%), PIK3CA (14.3%), and NRAS (5.7%) were the most frequent oncogene mutations, BRAF and FLT3 mutations occurred at lower frequencies, and mutations in the remaining oncogenes were rare. As shown in Table 4, this profile is consistent with the previous findings.^14–17

Table 4.

Summary of oncogene mutations in rectal cancer.

Gene	Mutation status in this study	Mutation status in Abdul-Jalil et al.’s study14	Mutation status in Gleeson et al.’s study15	Mutation status in Russo et al.’s study16	Mutation status in Russo et al.’s study17
KRAS	35.7% (25/70)	46.3% (93/201)	31.4% (32/102)	34.4% (22/64)	42.6% (20/47)
PIK3CA	14.3% (10/70)	14.4% (29/201)	6.9% (7/102)	12.5% (8/64)	4.3% (2/47)
NRAS	5.7% (4/70)	3.5% (7/201)	5.9% (6/102)	12.5% (8/64)	4.3% (2/47)
FLT3	2.9% (2/70)	No report	2.0% (2/102)	No report	0
BRAF	1.4% (1/70)	4.0% (8/201)	4.9% (5/102)	3.1% (2/64)	4.3% (2/47)
AKT1	0	1.5% (3/201)	0	0	No report
AKT2	0	0	No report	No report	No report
KIT	0	0	0	0	0
HRAS	0	No report	1.0% (1/102)	No report	No report
EGFR	0	1.0% (2/201)	0	0	0
MET	0	1.5% (3/201)	0	No report	No report
RET	0	0	0	No report	No report
PDGFRA	0	0.5% (1/201)	0	No report	No report
FGFR1	0	No report	0	No report	No report
FGFR3	0	No report	0	No report	No report
ABL1	0	No report	0	No report	No report
CDK	0	0	2.0% (2/102)	No report	No report
JAK2	0	No report	0	No report	0
ERBB2	0	No report	1% (1/102)	0	No report

The RAS-RAF-MAPK pathway has been associated not only with tumorigenesis but also with tumor progression. The RAS oncogene family includes KRAS, NRAS, and HRAS, which have been identified as biomarkers of anti-EGFR antibody treatment response in colorectal cancer.¹⁸ KRAS activation has been shown to contribute to increased intrinsic radiation resistance in human tumor cell lines,¹⁹ and several molecular mechanisms associating KRAS mutations and tumor radiation resistance have been described. Toulany et al.²⁰ showed that KRAS mutations upregulated tumor EGFR activity and downstream PI3K-AKT pathway via autocrine production and secretion of EGFR ligands to mediate radiation resistance. Likewise, Affolter et al.²¹ found that that the G12S KRAS mutation constitutively activated interaction of the PI3K-AKT and RAF-MAPK signaling pathways to enhance cellular survival after exposure to ionizing. Recently, Chakrabarti presented evidence that KRAS can reprogram glutamine metabolism in non–small-cell lung cancer (NSCLC) cell lines by driving the expression of the ME1 and GOT1 genes and interrupting redox balance to attenuate ionizing radiation–induced reactive oxygen species and DNA damage.²² In this series, patients with KRAS mutations had a higher probability of CRT resistance and worse long-term survival. Overall, KRAS was a valuable biomarker for predicting clinical efficacy to preoperative CRT in these LARC patients.

Previous studies have confirmed a clinical correlation of KRAS mutation subtypes and treatment outcomes in LARC. Abdul-Jalil et al.¹⁴ found that KRAS codon 12 (G12D, G12V, and G12S) and codon 13 (G13D) mutations were associated with a significantly worse DFS compared with KRAS wild-type tumors. Martellucci et al.²³ also found that LARC patients with KRAS codon 13 mutations were resistant to neoadjuvant CRT and less likely to achieve a pCR than patients without mutations. However, Lee et al.²⁴ found that KRAS mutation status was not associated with treatment response or postoperative prognosis in rectal cancer. That differs from our findings and might be attributed to different evaluation criteria. Lee et al. used tumor downstaging after CRT to assess response, but we employed TRG as the indicator. In addition, different KRAS mutation loci might be associated with different clinical outcomes after CRT treatment. Gaedcke et al.²⁵ reported that KRAS G12V mutations were associated with increased rates of tumor regression than G13D mutations (p = 0.012) and that patients with KRAS G12D mutations had worse 3-year OS than those without mutation at that locus (71.4% vs 95.1%, p = 0.028). KRAS mutations at different loci differ in ability to activate signal transduction pathways or may activate different pathways.^26,27 Specific KRAS mutations should be studied further to learn more of their specific association with distinct oncologic outcomes.

The oncogene mutation profiles identified in this study may have therapeutic and prognostic benefits for personalized cancer treatment. Genetic screening before preoperative CRT is necessary for LARC patients, as those with any RAS mutation had worse treatment outcomes. However, whether such patients should receive more intensive CRT regimens or directly undergo radical resection is yet to be investigated. In any case, more frequent follow-up is needed. We believe that a predictive system derived from the prognostic value of diverse oncogene mutations could be developed to guide precise treatment in LARC.

The phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3K)/AKT/mechanistic target of rapamycin (mTOR) pathway participates in tumor proliferation, differentiation, cell survival, trafficking, and glucose homeostasis and is one of the most important targets in anticancer therapy.^28,29 Mutations that are associated with activation of the PI3K pathway might be associated with poor survival outcome in rectal cancer.^30,31 However, PIK3CA gene mutations were not correlated with tumor response to preoperative CRT and survival in the LARC patients in this series; we found no significant correlation of tumor regression and BRAF mutation. The presence of an activating BRAF mutation in metastatic colorectal cancer has been associated with poor clinical outcome in patients treated with standard chemotherapy.^32,33 Nevertheless, the association of BRAF mutation and tumor response to CRT in LARC needs further investigation.

Conclusion

In summary, the most common oncogene mutations identified in these LARC patients were in KRAS, PIK3CA, and NRAS. KRAS mutation status influenced tumor response to preoperative CRT and survival in LARC. Further studies with larger sample sizes and longer follow-up are warranted to validate the clinical significance of these findings.

The authors would like to thank all the colleagues of Department of Colorectal Surgery in Sun Yat-sen University Cancer Center, who have involved with performing the treatment of the patients for this study. J.P., J.L., and M.Q. contributed equally to the study.

Declaration of conflicting interests

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authenticity of this article has been validated by uploading the key raw data onto the Research Data Deposit public platform (www.researchdata.org.cn), with the approval number as RDDB2017000059.

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. A waiver of informed consent was requested, and the approval was obtained from independent ethics committees at Sun Yat-sen University Cancer Center.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by grants from grant of Guangzhou Science and Technology Plan Projects (Health Medical Collaborative Innovation Program of Guangzhou) (grant no. 201400000001-4) and Sun Yat-sen University Clinical Research 5010 Program (2015024).