This article is mentioned in:

Introduction

Colorectal cancer (CRC) is one of the most common cancers worldwide, with more than one million newly diagnosed cases reported annually. In total, ~700,000 CRC-associated mortalities occurred worldwide in 2012, accounting for 8% of all cancer mortalities, and making CRC the fourth most common cause of cancer-associated mortality (1). The standard treatment for patients with unresectable metastases includes chemotherapy regimens based on irinotecan, oxaliplatin, fluoropyrimidines, anti-vascular endothelial growth factor (anti-VEGF) therapy (bevacizumab) and anti-epidermal growth factor receptor (anti-EGFR) therapy, such as panitumumab and cetuximab. These treatment options are efficient in a small percentage of patients, and it is important to identify specific biomarkers to determine the patients that are likely to benefit from anti-EGFR therapy (2–4).

EGFR triggers a downstream signalling cascade through, for example, the Kirsten rat sarcoma viral oncogene homolog-serine/threonine-protein kinase B-Raf and the phosphatidylinositol 3-kinase (PI3K), catalytic subunit α-phosphatase and tensin homolog-Akt pathways, which regulate cell proliferation, survival, apoptosis resistance, invasion and migration (2,3,5–8). The PIK3CA gene encodes the p110α subunit of PI3K α and belongs to class IA of the PI3Ks. The PI3K α protein is composed of regulatory subunit p85, which mediates anchorage to EGFR-specific docking sites, and catalytic subunit p110, which generates a second messenger that is responsible for the activation of Akt in response to the activation of growth factors from various ligands. These ligands include epidermal growth factor (EGF) or VEGF. Somatic mutations in cancer cells only occur in PIK3CA and PI3KR1, which encodes the p85 α subunit (3,6,9). These mutations are concentrated in two key regions of the PIK3CA gene, consisting of the helical domain of exon 9 and the kinase domain of exon 20 (7,8,10). Activating mutations in PIK3CA are detected in 7–32% patients, with G>A transversions in exon 9 being the most commonly observed configuration, which may coincide with KRAS and BRAF mutations. Tumours with the PIK3CA gene mutation are characterised by a predominant proximal colonic location (10,11) and by the frequent presence of mucinous differentiation (10).

Mutations of EGFR-dependent signalling molecules confer resistance to EGFR-specific antibody therapy. KRAS mutation is the first molecular marker of response to EGFR inhibitors (11). It has been hypothesised that the PIK3CA gene mutation may also affect the response to anti-EGFR therapy in patients with metastatic CRC (12,13). Certain studies indicate that PIK3CA exon 20 mutations negatively affect the response rate, disease control rate, progression-free survival (PFS) time and overall survival (OS) time, whilst PIK3CA exon 9 mutations demonstrate no significant effect on objective response (3,4).

The aim of the present study was to evaluate the importance of mutation in the PIK3CA gene as a prognostic factor in CRC. Additionally, the frequency of PIK3CA mutations in patients with CRC and the incidence of mutations in particular exons were examined. The association between the PIK3CA gene mutation and mutations in other downstream effectors of the EGFR signalling pathway was also analysed, in addition to the association between the PIK3CA gene mutation and various clinical or pathological features.

Materials and methods

Patient characteristics

Based on the database of the Military Institute of Medicine (Warsaw, Poland), 156 patients that were consecutively diagnosed with CRC were identified. The patients had been treated with palliative chemotherapy at the Oncology Department of the Military Institute of Medicine between 2006 and 2010. The inclusion criteria were as follows: Confirmed histopathological diagnosis of CRC; aged >18 years; presence of measurable lesions, determined by Response Evaluation Criteria in Solid Tumours, version 1.1 (14); adequate haematological parameters, consisting of a neutrophil count of ≥1.5×109, platelet count of ≥100×109/l and haemoglobin count of ≥9.0 g/dl; adequate biochemical parameters, comprising a bilirubin level <2 × upper limit of normal (ULN); an aspartate transaminase (AST); alanine transaminase (ALT) level <2.5 × ULN; a glomerular filtration rate (GFR) of >50 ml/min; and, in premenopausal women, an absence of pregnancy. The exclusion criteria were as follows: Renal insufficiency, demonstrated by a GFR <50 ml/min; hepatic insufficiency, demonstrated by AST and ALT levels >2.5 × ULN; and severe concomitant disease, such as unstable cardiac angina. This study was approved by the ethics commmittee of the Military Institute of Medicine and written informed consent was obtained from all patients.

Histopathological examination of tumour specimens

Primary tumour specimens were collected from CRC patients. Formalin-fixed paraffin embedded (FFPE) tissue blocks were cut into serial 5 µm-thick sections for haematoxylin and eosin staining. The presence of tumour tissue was verified by an experienced pathologist. Subsequently, tissue samples from at least three serial sections were macrodissected to ensure that the specimens contained ≥80% tumour cells.

DNA extraction

DNA from FFPE colorectal tumour tissues was isolated from 10–30-µm thick sections subsequent to macrodissection, resulting in the selection of specimens containing ≥80% tumour cells. Tumour samples were extracted with xylene and ethanol to remove paraffin, and placed in 1% SDS/proteinase K (10 mg/ml) at 56°C overnight. DNA was isolated using the NucliSENS easyMag platform (bioMérieux, Marcy l'Etoile, France) for automated nucleic acid extraction.

KRAS, BRAF and PIK3CA mutation analysis

The detection of mutations in codons 12 and 13 of exon 1 of the KRAS gene and exons 11 and 15 of the BRAF gene was conducted using a previously described method (15). The analysis of mutations in exons 9 and 20 of the PIK3CA gene was performed by direct sequencing, as described by Samuels et al (16) and Li et al (17), with a number of modifications. The primers for exon 9 were designed to avoid amplification of homologous sequences located at the chromosome 22q11.2 cat-eye syndrome region and on chromosome 16. Genomic DNA obtained from tumour samples was amplified by polymerase chain reaction (PCR) using the following primers: Forward strand exon (FSE)9, 5′-TTGCTTTTTCTGTAAATCATCTGTG-3′; Reverse strand exon (RSE)9, for exon 9 of PIK3CA, 5′-CTGCTTTATTTATTCCAATAGGTATG-3′; FSE20, 5′-ACATCATTTGCTCCAAACTGA-3′, RSE20, for exon 20 of PIK3CA, 5′-CATAACATGAAATTGCGCATT-3′. PCR was conducted in a total volume of 10 µl, containing 2 µl of the extracted genomic DNA, using 10X PCR buffer, 1.5 mmol/l MgCl2, 0.2 µmol/l of each primer, 0.1 mmol/l of deoxynucleoside triphosphate, and 1 unit of Taq DNA polymerase (Eurx Ltd., Gdańsk, Poland). PCR conditions were as follows: 95°C for 10 min; 45 cycles of 95°C for 30 sec, 59°C for 30 sec and 72°C for 30 sec; and finally 7 min at 72°C. The amplification products were purified using the DNA Gel-Out kit (DNA Gdańsk, Gdynia, Poland). Automated sequencing was conducted using the BigDye® Terminator v3.1 Cycle Sequencing Kit (Life Technologies, Warsaw, Poland). Sequencing reactions were purified using the ExTerminator kit (DNA Gdańsk), and analysed on a 3500 Genetic Analyzer sequencer (Life Technologies). A wild-type control DNA sample without PIK3CA mutation and a known mutation sample (substitution 1633 G>A, E545K within exon 9 and substitution 3140 A>G, H1047R within exon 20) were included in the experiment. The mutation was confirmed by sequencing at least two independent PCR products.

Statistical analysis

The OS time was defined as the time elapsed between the commencement of the first line of palliative chemotherapy, and the date of mortality or of the final follow-up, and was estimated according to the Kaplan-Meier method. The cut-off date for the present analysis was December 2013.

The χ2 test was used to investigate the association between variables in the two treatment groups with respect to baseline characteristics. The log-rank test was performed in the Kaplan-Meier survival analyses to assess differences between the groups with regard to OS time. P<0.05 was considered to indicate a statistically significant difference. Multivariate analyses of OS time were performed by Cox proportional-hazard regression using the forward stepwise method, all variables determined to be significant in the univariate analysis were included in the multivariate analysis. Analyses were performed using the statistical package Statistica, version 7.0 (Statsoft, Inc., Tulsa, OK, USA).

Results

Patient characteristics

The characteristics of included patients are summarised in Table I. The cohort comprised 100 women and 56 men, with a median age of 67 years. The majority of patients (91%) underwent primary tumour resection. The primary tumour was located in the colon in 67 patients (42.9%), and in the sigmoid colon or in the rectum in 89 patients (57.1%). Metastases were located in the liver in 77 patients (70.6% of patients with metastases), the lungs in 21 patients (19.3% of patients with metastases), and other organs in 72 patients (66.0% of patients with metastases). Lymph node metastasis was also identified in 50.6% of patients. The majority of patients had a good performance status (Karnofsky status of 80–100).

Table I. Characteristics of the patients with colorectal cancer enrolled in the present study.

Table I.

Characteristics of the patients with colorectal cancer enrolled in the present study.

Characteristic	Value
Total, n	156
Age in years, median (range)	67 (25–85)
Gender, n (%)
Female	100 (64.9)
Male	56 (35.9)
KRAS status, n (%)
Mutation	44 (28.2)
Codon 12	41 (26.3)
Codon 13	3 (1.9)
Wild-type	112 (71.8)
BRAF status, n (%)
Mutation	12 (7.7)
Wild-type	144 (92.3)
PIK3CA status, n (%)
Mutation	15 (9.6)
Codon 9	7 (4.5)
Codon 20	8 (5.1)
Wild-type	141 (90.4)
Primary tumour localisation, n (%)
Colon	67 (42.9)
Sigmoid/rectum	89 (57.1)
Localisation of metastases (n=109), n (%)
Liver	77 (70.6)
Lungs	21 (19.3)
Other localisations	72 (66.0)
Karnofsky performance status, n (%)
100	79 (50.6)
90	63 (40.4)
80	12 (7.7)
70	2 (1.3)
Histological differentiation grade, n (%)
High/moderate	126 (80.8)
Low/unknown	30 (19.2)
Histological type, n (%)
Mucinous	7 (4.5)
Mixed	44 (28.2)
Cylindocellular	3 (1.9)
Tubular	70 (44.9)
Unclassified	32 (20.5)
Previous adjuvant chemotherapy, n (%)	55 (35.3)
Lymph node status, n (%)
N0	35 (22.4)
N1	42 (26.9)
N2a	20 (12.8)
N2b	17 (10.9)
Nx	42 (27.0)
Invasive extent, n (%)
Tx	10 (6.4)
T1	1 (0.6)
T2	17 (10.9)
T3	105 (67.3)
T4	23 (14.7)

[i] N, lymph node involvement stage; T, tumour stage.

KRAS, BRAF, PIK3CA gene mutation status

KRAS mutations were present in 44 patients (28.2%), of whom 41 patients (93.2%) possessed mutations in codon 12, and three patients (6.8%) possessed mutations in codon 13. BRAF mutations were present in 12 patients (7.7%). PIK3CA mutations were present in 15 patients (9.6%), of whom seven (46.7%) had mutations in codon 9, and eight (53.3%) had mutations in codon 20. Mutation in the PIK3CA gene was detected in six patients who had KRAS gene mutations (40% of PIK3CA mutated tumours) and in one patient with BRAF mutations (6.6% of PIK3CA mutated tumours).

Clinicopathological variables and PIK3CA gene mutation status

The evaluation of PIK3CA mutation status relative to clinicopathological variables is summarised in Table II. PIK3CA gene mutations were present in 15 patients (9.6%). An increased incidence of PIK3CA gene mutations was detected in patients with involved lymph nodes, with low-grade or unknown histological differentiation of the tumour, and with tubular cancer. PIK3CA gene mutations were also frequently present in patients with advanced disease, T stage III or IV. However, the association between these variables and PIK3CA status was not statistically significant.

Table II. Comparison of PIK3CA gene mutation status and clinicopathological variables (n=156).

Table II.

Comparison of PIK3CA gene mutation status and clinicopathological variables (n=156).

	PIK3CA status
Clinicopathological variable	Wild-type	Mutation	Statistical test	P-value
Patients, n	141	15
Age, years			1026.0a	0.8521
Median	67	69
Range	25–85	37–79
Gender, n			0.25c	0.6165
Male	52	4
Female	89	11
Histological differentiation grade, n			2.70c	0.1003
High/moderate	30	0
Low/unknown	111	15
Primary tumour localisation, n			0.07b	0.7914
Sigmoid/rectum	79	9
Colon	61	6
Karnofsky performance status, n			0.65c	0.4214
≤80	14	0
>80	127	15
Primary tumour size, n			0.04c	0.8445
T1/T2	17	1
T3/T4	124	14
Lymph node involvement grade, n			2.92b	0.0873
N0	29	6
N positive	112	9
Histological type, n			0.02b	0.8835
Tubular	63	7
Other	78	8
BRAF status, n			0.12c	0.7242
Wild-type	130	14
Mutation	11	1
KRAS status, n			1.13b	0.2872
Wild-type	103	9
Mutation	38	6

a Mann-Whitney test

b V-square

c Yates' χ² test. T, tumour stage; N0, no lymph node involvement; N positive, lymph node involvement.

Prognostic significance of PIK3CA gene mutation status

No significant difference in OS rate was identified between patients with PIK3CA mutations and those with wild-type PIK3CA genes (P=0.1270; Fig. 1). However, patients with PIK3CA mutations tended to demonstrate a decreased OS rate. The median OS in patients with wild-type PIK3CA genes was 56.7 months, compared with 47.6 months in patients presenting with mutations.

Figure 1. OS rates in patients with PIK3CA gene mutations, relative to those with wild-type PIK3CA. OS, overall survival.

Clinical and pathological variables identified by univariate analysis as potential prognostic factors for OS rate

The results of the univariate analysis are summarised in Table III. Univariate analysis of the present patient cohort identified that gender and lymph node involvement acted as prognostic factors that influenced OS rate, as female patients survived for 57.5 months, compared with 39.3 months for male patients (P=0.0111, and the median OS in patients without lymph node metastases was 61.4 months, compared with 45.4 months in patients presenting with metastases (P=0.0122). The OS rate associated with other clinical parameters, consisting of age, primary tumour localisation, Karnofsky performance status, histological type, histological differentiation grade, primary tumour size and gene mutation status, did not differ significantly between groups.

Table III. Univariate analysis of OS rate (log-rank test).

Table III.

Univariate analysis of OS rate (log-rank test).

Clinical parameter	n	Median OS, months	P-value
Age, years			0.9269
<70	102	59.2
≥70	54	45.4
Gender			0.0111a
Male	56	39.3
Female	100	57.5
Primary tumour localisation			0.9432
Sigmoid/rectum	67	44.6
Colon	89	49.6
Karnofsky performance status			0.6373
≤80	14	20.9
>80	142	49.4
Lymph node involvement			0.0122a
Present	121	45.4
Absent	35	61.4
Histological type			0.9808
Tubular	86	48.8
Other	70	47.8
Histological differentiation grade			0.1331
High/moderate	126	52.4
Low/unknown	30	29.3
Primary tumour size			0.1280
T1/T2	18	60.1
T3/T4	138	47.4
PIK3CA status			0.1271
Mutation	15	56.7
Wild-type	141	47.6
KRAS status			0.7740
Mutation	44	47.9
Wild-type	112	48.8
BRAF status			0.6398
Mutation	12	22.7
Wild-type	144	49.4

a Statistically significant correlation (P<0.05). OS, overall survival; T, tumour stage.

Clinical and pathological variables identified by multivariate analysis as potential prognostic factors for OS rate

The results of the multivariate analysis are summarised in Table IV. Multivariate analysis identified that lymph node involvement grade [hazard ratio (HR), 1.68; P=0.0467] and male gender (HR, 1.57; P=0.0249) were adverse prognostic factors for OS rates. KRAS, BRAF and PIK3CA gene mutation status was not found to significantly affect OS rate in this analysis.

Table IV. Multivariate analysis of overall survival rate.

Table IV.

Multivariate analysis of overall survival rate.

Clinical parameter comparison	Multivariate analysis, HR (95% CI)	P-value
Lymph node involvement	1.68 (1.01–2.82)	0.0467
Gender	1.57 (1.06–2.32)	0.0249
PIK3CA status	NS	>0.05
KRAS status	NS	>0.05
BRAF status	NS	>0.05

[i] NS, not significant; HR, hazard ratio; CI, confidence interval.

Discussion

The treatment of cancer is increasingly based on targeted therapy, including morphological identification of tumour histology, tumour staging and identification of target pathways and molecules. It has been established that KRAS mutation is a marker of resistance to anti-EGFR therapy in patients with CRC (4,18–21). Despite the exclusion of patients with KRAS-mutant tumours, anti-EGFR treatment fails in numerous patients with CRC. A number of studies have demonstrated a negative correlation between PIK3CA mutations and clinical outcomes (3,4,8,10,12,22).

The aim of the current study was to evaluate the incidence of PIK3CA gene mutation in patients with CRC at all stages, and also to determine the association between mutation of the PIK3CA gene and mutations in other downstream effectors of the EGFR signalling pathway. Additionally, the incidence of mutations in exon 9 and exon 20 of the PIK3CA gene were examined. The present study also evaluated the role of the PIK3CA gene mutation and the select clinical and pathological variables of these tumours as potential prognostic factors.

Activating mutations in the PIK3CA gene are identified in 7–32% of CRC patients. In the present study, PIK3CA mutations were detected in 9.6% of CRCs, which is consistent with previously published data (3,7,8,10–12,18,23,24). However, in contrast to a number of earlier studies (3,10,11,18,23–25), the present analysis identified a similar frequency of mutations in exons 9 and 20. Overall, 46.7% of patients possessed mutations in codon 9, while 53.3% possessed mutations in codon 20. Pentheroudakis et al (25) detected the PIK3CA mutation in exon 9 in 54% of cases, and in exon 20 in 13.5% of cases. Double PIK3CA gene mutations in exons 9 or 20 were not detected in the present cohort, in contrast to the studies conducted by Rock et al (18) and Sartore-Bianchi et al (26). In the present study, mutation in the PIK3CA gene coincided with KRAS gene mutations in six patients, comprising 40% of PIK3CA mutated tumours, and with BRAF mutations in one patient, comprising 6.6% of PIK3CA mutated tumours. Similar associations have been reported in previous studies (10,12,24,26).

Tumours with PIK3CA mutations were characterised by a predominantly distal colonic location, the frequent presence of tubular differentiation and a low grade of histological differentiation. These results are in contrast with previously published studies, with certain studies identifying no clinical features that were associated with PIK3CA gene mutations (7,22), while others identified an association between PIK3CA gene mutation and tumour mucinous differentiation and proximal colon location (10,24). Notably, patients with more advanced disease, T stage III or IV, or those demonstrating the involvement of lymph nodes presented with an increased rate of PIK3CA gene mutations. Similar associations have been previously reported (7,27). The difference in the clinicopathological characteristics of the tumour between the mutation statuses of particular exons was not estimated due to the small number of tumours demonstrating PIK3CA mutations in the present patient population.

In the present study, the results of the univariate and multivariate analyses into the role of clinical and pathological variables revealed a positive, statistically significant association between female gender and uninvolved lymph nodes on the overall patient survival.

The present study did not confirm a prognostic role for PIK3CA mutation status in CRC patients, in contrast to the results obtained by Rosty et al (10) and Therkildsen et al (28), who observed a shorter survival time in patients with PIK3CA mutations. The current results are consistent with findings reported by Cathomas (24), Zhu et al (7) and Karapetis et al (29), that PIK3CA exhibited no prognostic impact. Ogino et al (30) also reported that tumour PIK3CA mutation status is not associated with stage III colon cancer prognosis. Compared with carriers of wild-type PIK3CA, patients with a PIK3CA-mutated tumour had a shorter OS rate. However, this trend was not statistically significant. Similar data have been reported in previous studies (11,18,26). Studies have reported differences between exon 9 and 20 mutations with regard to their effects on PFS and OS, noting that PIK3CA exon 20 mutations were significantly associated with poorer PFS and OS (3,4). The biological effects of mutations in exons 9 and 20 of the PIK3CA gene were not compared in the present study due to the small number of patients with mutant PIK3CA. It has previously been reported that the coexistence of PIK3CA exon 9 and 20 mutations is associated with poor prognosis in CRC patients (31).

The results of the present study indicate that aberrations in PIK3CA did not contribute significant prognostic information. The role of the PIK3CA mutation status remains unclear; therefore future prospective multi-centre trials involving CRC patients are essential in order to fully assess the clinical relevance of the PIK3CA mutation status.

In summary, activating mutations in the PIK3CA gene were present in 9.6% of colorectal carcinomas, and coincided with mutations in other downstream effectors of the EGFR signalling pathway. The results from this analysis of CRC patients of all disease stages indicates that the PIK3CA mutation status is not a prognostic factor in these patients. In addition, there is no statistically significant association between PIK3CA mutation and clinicopathological factors.

References