PIK3CA Mutation, Aspirin Use after Diagnosis and Survival of Colorectal Cancer. A Systematic Review and Meta-analysis of Epidemiological Studies

Clinical Oncology xxx (2015) 1e10 Contents lists available at ScienceDirect

Clinical Oncology journal homepage: www.clinicaloncologyonline.net

Original Article

PIK3CA Mutation, Aspirin Use after Diagnosis and Survival of Colorectal Cancer. A Systematic Review and Meta-analysis of Epidemiological Studies L. Paleari *y, M. Puntoni z, M. Clavarezza *, M. DeCensi *, J. Cuzick x, A. DeCensi *x{ * Division

of Medical Oncology, E.O. Ospedali Galliera, Genoa, Italy Public Health Agency, Liguria Region, Italy z Office of the Scientific Director, E.O. Ospedali Galliera, Genoa, Italy x Wolfson Institute of Preventive Medicine, Queen Mary University of London, UK { Division of Cancer Prevention and Genetics, European Institute of Oncology, Milan, Italy y

Received 23 July 2015; received in revised form 11 October 2015; accepted 13 October 2015

Abstract Aims: Regular aspirin use has been associated with inhibition of the whole spectrum of colorectal carcinogenesis, including prevention of metastases and reduced total mortality in colorectal cancer. Preclinical data show that aspirin down-regulates PI3 kinase (PI3K) signalling activity through cyclo-oxygenase-2 (COX-2) inhibition, leading to the hypothesis that the effect of aspirin might be different according to PIK3CA mutational status, but epidemiological studies have led to conflicting results. The aim of this study was to assess the relationship between PIK3CA status and the efficacy of regular use of aspirin after diagnosis on overall survival in colorectal cancer patients. Materials and methods: We identified studies that compared post-diagnosis aspirin efficacy in colorectal cancer patients identified by PIK3CA status. Hazard ratios for overall survival were meta-analysed according to PIK3CA status by inverse variance weighting. A pooled test for treatment by PIK3CA status interaction was carried out by weighted linear meta-regression. All statistical tests were two-sided. Results: The overall effect of aspirin was not significant (summary risk estimate ¼ 0.82; 95% confidence interval 0.63e1.08, P ¼ 0.16; I2 ¼ 57%). In PIK3CA mutant disease (n ¼ 588), aspirin use reduced total mortality by 29% (summary risk estimate ¼ 0.71; 95% confidence interval 0.51e0.99, P ¼ 0.04; I2 ¼ 0%), whereas in PIK3CA wild-type disease (n ¼ 4001), aspirin use did not reduce overall mortality (summary risk estimate ¼ 0.93; 95% confidence interval 0.61e1.40; P ¼ 0.7; I2 ¼ 80%) (P interaction ¼ 0.39). There was a beneficial trend for aspirin on cancer-specific survival in PI3KCA mutated subjects (summary risk estimate ¼ 0.37, 95% confidence interval 0.11e1.32, P ¼ 0.1), albeit with high heterogeneity (Q chi-squared ¼ 3.41, P ¼ 0.07, I2 ¼ 70.7%). Conclusion: These findings suggest that the benefit of post-diagnosis aspirin treatment on overall mortality in colorectal cancer may be more marked in PIK3CA mutated tumours, although the low number of studies prevents definitive conclusions. Trials addressing this issue are warranted to assess the efficacy of aspirin in the adjuvant setting. Ó 2015 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.

Key words: Adjuvant trials; aspirin; colorectal cancer; overall survival; PIK3CA; post-diagnosis use

Introduction Death from cancer is almost always from metastatic disease. A recent meta-analysis of five large cardiovascular

Author for correspondence: A. DeCensi, Division of Medical Oncology, E.O. Ospedali Galliera, Mura delle Cappuccine 14, 16128 Genoa, Italy. Tel: þ39-0105634501; Fax: þ39-01057481090. E-mail address: [email protected] (A. DeCensi).

trials involving over 17 000 patients provided proof-ofprinciple that aspirin can reduce cancer incidence and the development of metastasis [1]. Specifically, patients with colorectal cancer (CRC) who received aspirin before and during the study period considerably reduced their risk of developing subsequent metastasis (hazard ratio 0.26, 95% confidence interval 0.11e0.57) with a 49% relative reduced risk of death (hazard ratio 0.61, 95% confidence interval 0.47e0.80) and a greatest benefit for cancer

http://dx.doi.org/10.1016/j.clon.2015.11.008 0936-6555/Ó 2015 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.

Please cite this article in press as: Paleari L, et al., PIK3CA Mutation, Aspirin Use after Diagnosis and Survival of Colorectal Cancer. A Systematic Review and Meta-analysis of Epidemiological Studies, Clinical Oncology (2015), http://dx.doi.org/10.1016/j.clon.2015.11.008

2

L. Paleari et al. / Clinical Oncology xxx (2015) 1e10

at the proximal colon (hazard ratio ¼ 0.35, 95% confidence interval 0.20e0.63) [1e3]. Moreover, recent observational and randomised controlled studies in cardiovascular disease have shown that regular use of aspirin after CRC diagnosis may prolong overall survival [4e8], suggesting its assessment in the adjuvant therapy of CRC, but also highlighting the notion that CRC is a heterogeneous group of diseases with potentially different responses to aspirin treatment. The putative effects of aspirin on distant metastasis are consistent with experimental evidence in animals, where platelets play an important role in blood-borne metastases, which is already critical after the first days or weeks of the carcinogenesis progression [9]. Platelets aggregate the circulating cancer cells, masking the cytolytic effect of natural killer cells [9], thus allowing adhesion to macrophages, which in turn favour the invasion of the endothelial cells by the cancer cells [9e12]. Furthermore, inflammation, an accepted hallmark of cancer [13], plays a critical role in CRC development and prognosis [14,15]. One of the main mechanisms of action of aspirin is the inhibition of cyclo-oxygenase-2 (COX-2 encoded by PTGS2), the enzyme responsible for the biosynthesis of prostaglandins, which enhance cellular proliferation, migration, invasiveness and tumoral neoangiogenesis promotion [16e18]. It is experimentally proven that alterations in PTGS2 expression and abundance of its enzymatic product prostaglandin E2 play a central role in influencing the development of CRC [19], although this takes place at a much higher dose than used for cardiovascular prevention. In addition, COX-2 overexpression has been associated with worse survival among CRC patients [16e18]. Despite this growing body of evidence, the precise biological mechanisms underlying the antineoplastic effect of aspirin in CRC are still unknown. The identification of biomarkers able to predict response to aspirin would be of considerable clinical value towards a precision medicine approach. A recent epidemiological study suggested that the benefit of regular use of aspirin after CRC diagnosis on overall survival was restricted to patients whose tumours harbour PIK3CA mutations, although exploratory analyses suggested the need for external validation before definitive conclusions [4]. The PIK3CA gene encodes the p110a catalytic subunit of PI3 kinase (PI3K) whose signalling pathway plays an important role in human carcinogenesis [20]. PIK3CA mutations result in constitutive activation of PI3K and the downstream AKT pathway, which enhance PTGS2 activity and prostaglandin E2 synthesis, resulting in inhibition of apoptosis in CRC cells [21,22] and COX-2 upregulation [23]. Aspirin may therefore suppress cancer cell growth and induce apoptosis by blocking the PI3K pathway and this effect might differ according to PIK3CA tumour status [4e8]. We carried out a systematic review and meta-analysis of non-randomised studies to assess the relationship between PIK3CA status and the efficacy of regular use of aspirin after diagnosis on overall survival in CRC patients.

Materials and Methods This review was conducted following the guidelines in the Meta-analysis of Observational Studies in Epidemiology (MOOSE) and the PRISMA statement for reporting of systematic reviews [24,25]. Search Strategy We identified studies by searching Medline, ISI Web Science (Science Citation Index Expanded), Embase, Cochrane Library, by examining the reference list of published trials, reviews articles and editorials and by hand-searching references in identified trials and symposia reports from the major cancer associations. Any study that examined the relationship between the use of aspirin and PIK3CA status in CRC patients was eligible for inclusion in our review. For the database search we used [((‘Aspirin’ [Mesh]) OR ‘aspirin’) AND (‘PIK3CA protein, human’ [Supplementary Concept] OR ‘PI3KCA protein, human [Supplementary Concept]’ OR ‘pik3ca’ OR ‘pi3kca’)] as the search terms. The literature search was independently carried out by two reviewers (LP and MP) with a standardised approach and discrepancies were resolved by consensus. The databases were searched for papers published to 31 January 2015. Criteria for Inclusion All the studies considered for this meta-analysis had to fulfil the following criteria: (i) to report the association between the post-diagnosis use of aspirin among CRC patients and overall survival according to PIK3CA mutational status; (ii) to report effect estimates of aspirin use and 95% confidence intervals; (iii) to be independent studies. All English-language published resources (as full paper or conference abstract) were eligible for inclusion in our analyses. In case of duplicate reports of the same study, only the most recent one was included. Data Extraction The same reviewers who undertook the literature search extracted the data independently using standardised data collection forms. Data retrieved from the reports included publication details, methodological components, study characteristics, such as sample size, interventions, followup duration, outcome measures, effect estimates with 95% confidence interval and covariates in the fully adjusted model. All data were checked for internal consistency and the authors of the studies were contacted, if needed, to recover missing information, update outcome measures or clarify inconsistencies. Statistical Analysis The primary end point of this analysis was overall survival, given the lack of information on cancer-specific survival (CSS). Hazard ratios or relative risk estimates

Please cite this article in press as: Paleari L, et al., PIK3CA Mutation, Aspirin Use after Diagnosis and Survival of Colorectal Cancer. A Systematic Review and Meta-analysis of Epidemiological Studies, Clinical Oncology (2015), http://dx.doi.org/10.1016/j.clon.2015.11.008

L. Paleari et al. / Clinical Oncology xxx (2015) 1e10

for mortality were used as treatment effect estimates and computed setting aspirin non-users as the reference group. All data were checked for internal consistency; the principal investigators were contacted to retrieve missing information or to solve inconsistencies. Weighted summary risk estimates (SRE) were calculated by pooling effect estimates across the studies by inverse variance weighting. A value lower than 1 indicates a protective effect of aspirin. Because all studies included in the main analysis had different clinical characteristics, we examined the extent of heterogeneity among them visually, by means of Forest-plots, using Cochran’s Q test, and the Isquared (I2) parameter, which represents the percentage of total variation across studies that is attributable to heterogeneity rather than to chance. Cochran’s Q test P values  0.10 were considered significant for heterogeneity and I2 < 25% was considered as an acceptable (low) level of heterogeneity. The general approach of the analysis was to primarily apply random effects modelling using the method of DerSimonian and Laird [26], with the estimate of heterogeneity being taken from the Mantel-Haenszel model. Sensitivity analyses were applied to evaluate the extent to which the combined estimate might be affected by individual studies (leave-one-out procedure) and considering a putative source of heterogeneity such as PIK3CA mutation method assessment. We did not conduct further statistical tests for funnel plot asymmetry because of the limited test power when fewer than 10 studies are included [27]. We used meta-regression modelling to test the influence of PIK3CA mutational status on the association between the post-diagnosis use of aspirin among CRC patients and overall survival. This analysis, which can be considered equivalent to a test for interaction when individual patient data are not available, was conducted by testing the difference among the pooled estimates in the subgroups of studies (PIK3CA mutated versus wild-type subjects). The regression coefficient indicates how the intervention effect in each subgroup differs from the effect in a specified reference subgroup and the P value of the regression coefficient indicates whether this difference is statistically significant. All analyses were accomplished using STATA (version 13.1, StataCorp LP, College Station, TX, USA). All P values were two-sided and the cut-off for statistical significance was 0.05.

Results We identified and retrieved five studies including 4589 CRC patients together with aggregated data on aspirin use after diagnosis and the presence (or absence) of PIK3CA mutation by the search strategy summarised in Figure 1. Four studies were published as full articles in peer-review journals [4,6e8] and one was available only as a conference abstract [5].

3

Fig 1. Flow diagram of the study search and selection.

Characteristics of Included Studies The five studies were published between 2012 and 2014. Two studies were conducted in the USA, one in Australia and the USA, one in the UK and one in the Netherlands. The sample size of the included studies ranged from 243 to 1487. One study enrolled patients with colon cancer only [8], whereas the remaining studies included both colon and rectal cancer [4e7]. Two of the included studies examined patients with CRC at all stages (IeIV) [4,8], one study reported IIeIII stage patients [6], one enrolled stage IIeIV patients [7] and one did not report stage information [5]. All studies reported the association between the use of aspirin in accordance with PIK3CA mutational status and overall survival; two studies also reported estimates for CSS [4,7] and one for recurrence-free survival [6]. The characteristics of each study are listed in Table 1. Table 2 shows PIK3CA testing methodology and scoring criteria in individual studies. Overall, the proportion of patients who had the PIK3CA status of their CRC assessed was 92% (n ¼ 4221/4589). Among the patients whose PIK3CA status was assessed (n ¼ 4221), 588 (14%) had a mutated PIK3CA gene. PIK3CA status was evaluated by Sanger sequencing in two studies [6,8], pyrosequencing in one study [4], not reported in one study [5] and by Sanger

Please cite this article in press as: Paleari L, et al., PIK3CA Mutation, Aspirin Use after Diagnosis and Survival of Colorectal Cancer. A Systematic Review and Meta-analysis of Epidemiological Studies, Clinical Oncology (2015), http://dx.doi.org/10.1016/j.clon.2015.11.008

4

Reference, country

Study design

End point(s)

[4], USA and Japan

Prospective OS, CSS cohort studies*

PIK3CA mutated aspirin users/total sample size

Aspirin Disease stage, dose used n (%) daily

66/964

325 mg

Median follow-up

Risk estimates (95% confidence interval)

CSS: Stage I: 260 (27) 153 months Stage II: 301 (31) (IQR: 104e195) e mutated PIK3CA Stage III: 264 (27) HR ¼ 0.18 Stage IV: 64 (7) (0.06e0.61); Unknown: 75 (8) e wild-type PIK3CA HR ¼ 0.96 (0.69e1.32); OS: e mutated

PIK3CA HR ¼ 0.54 (0.31e0.94); e wild-type PIK3CA HR ¼ 0.94 (0.75e1.17);

Adjusting covariates (in addition to age and gender) Disease stage, year of diagnosis, time of aspirin use after diagnosis, use of aspirin before diagnosis, tumour location, tumour differentiation, BMI, microsatellite instability status, cpg island methylator phenotype, KRAS mutation, BRAF mutation, LINE-1 methylation, PTGS2 expression

Interaction: P ¼ 0.009 (CSS), P ¼ 0.07 (OS)

[5], USA (Texas)

Early phase clinical trials

OS

25/243

N.R.

N.R.

N.R.

[6], UK

Prospective cohort study nested in a randomised trial (VICTOR trial)

RFS, OS

14/896

100 mg

Stage II: 443 (49) 62 months Stage III: 453 (51) (IQR: 50e70) No stage III/IV

OS HR ¼ 1.37 (0.80e2.35) CRC wild-type PIK3CA, HR ¼ 1.80 (1.01e3.23) CRC mutated PIK3CA, HR ¼ 0.75 (0.17e3.20)

RFS: e mutated PIK3CA HR ¼ 0.11 (0.001e0.832); e wild-type PIK3CA HR ¼ 0.94 (0.59e1.49);

Royal Marsden Hospital score: albumin <3.5 versus 3.5 g/dl, lactate dehydrogenase > upper limit of normal (ULN) versus  ULN, and >2 versus 2 sites of metastases. Disease stage, tumour location, tumour grade, microsatellite instability, prior chemotherapy or radiotherapy

L. Paleari et al. / Clinical Oncology xxx (2015) 1e10

Please cite this article in press as: Paleari L, et al., PIK3CA Mutation, Aspirin Use after Diagnosis and Survival of Colorectal Cancer. A Systematic Review and Meta-analysis of Epidemiological Studies, Clinical Oncology (2015), http://dx.doi.org/10.1016/j.clon.2015.11.008

Table 1 Epidemiological studies of post-diagnosis aspirin use in PIK3CA mutated and wild-type colorectal cancer (CRC)

PIK3CA HR: 0.29 (0.04e2.33); e wild-type PIK3CA HR ¼ 0.95 (0.56e1.61); Interaction: P ¼ 0.024 (RFS), P ¼ 0.26 (OS).

[7]y, USA (Florida) and Australia

Prospective cohort study

[8], the Netherlands Retrospective cohort study

OS, CSS, RFS 49/1487

85%  100 Stage mg Stage Stage Stage

OS

N.R.

27/999

I: 8 (4) II: 66 (36) III: 67 (36) IV: 44 (24)

54 months (IQR N.R.)

Stage I: 138 (14) 84 months Stage II: 402 (40) (IQR N.R.) Stage III: 287 (29) Stage IV: 169 (17) Unknown: 3 (0)

OS HR ¼ 0.95 (0.55e1.63), P ¼ 0.85; CSS HR ¼ 0.66 (0.31e1.38), P ¼ 0.27. Overall RR ¼ 0.64 (0.49e0.83) e PIK3CA mutated: RR ¼ 0.73 (0.33e1.63)

Aspirin, stage, cancer centre and primary site

Pathologic/clinical stage, adjuvant chemotherapy, comorbidity, tumour grade, and year of diagnosis

[9 events/27 aspirin users] e PIK3CA

wild-type: RR ¼ 0.55 (0.40e0.75) [55 events/147 aspirin users]

N.R., not reported; IQR, interquartile range; OS, overall survival; RFS, relapse-free survival; CSS, cancer-specific survival; BMI, body mass index; HR, hazard ratio; RR, relative risk. * Data from the Nurses’ Health Study and the Health Professionals Follow-up Study. y All data refer to PIK3CA mutated tumours only (n ¼ 185). All estimates come from multivariate analyses, models are adjusted for the covariates indicated.

L. Paleari et al. / Clinical Oncology xxx (2015) 1e10 5

Please cite this article in press as: Paleari L, et al., PIK3CA Mutation, Aspirin Use after Diagnosis and Survival of Colorectal Cancer. A Systematic Review and Meta-analysis of Epidemiological Studies, Clinical Oncology (2015), http://dx.doi.org/10.1016/j.clon.2015.11.008

OS: e mutated

6

L. Paleari et al. / Clinical Oncology xxx (2015) 1e10

Table 2 Assessment methods of PIK3CA status in the studies included in the meta-analysis Reference

PIK3CA status No. patients assessed /no. patients (%)

PIK3CA assay method No. of patients PIK3CA mutant/ no. of patients assessed (%)

[4] [5] [6]

964/964 (100%) 243/243 (100%) 896/896 (100%)

161/964 (16.7%) 38/242 (15.6%) 104/896 (11.6%)

[7]

1487/1487 (100%)

185/1487 (12.4%)

[8]

631/999 (63.16%)

100/631 (15.8%)

Central testing

Type of assay

Not reported Not Reported Not reported

Pyrosequencing (exons 9 and 20) Not Reported Sanger sequencing [exon 9 (codons 513e554); exon 20 (codons 992e1068)] Sanger sequencing (exons 9 and 20) for RMH patients Next Generation Sequencing for MCC patients Hydrolysis probe assay for the hotspot mutations Sanger sequencing for additional non-hotspot mutations (exons 9 and 20)

No

Not Reported

RMH, Royal Melbourne Hospital; MCC, Moffitt Cancer Canter.

sequencing or Illumina Next Generation Sequencing according to two different cohorts in one study [7]. Despite this analytical heterogeneity, the prevalence of mutated PIK3CA tumours was quite constant among the five studies, ranging from 12 to 17%.

Sensitivity analyses of single study influence on the combined effect estimates (leave-one-out study procedure) and the adoption of two different detection assays for PIK3CA status did not significantly change the SRE (data not shown).

Discussion Effect of Aspirin on Overall Survival and the Effect of PIK3CA Mutational Status Data on overall survival were available in all the studies included in this meta-analysis. Globally, post-diagnosis aspirin use was associated with a non-significant trend for a reduction in overall mortality compared with non-use (SRE ¼ 0.82, 95% confidence interval 0.63e1.08, P ¼ 0.16) (Figure 2), with a significant heterogeneity between studies (Cochran’s Q chi-square ¼ 18.63, P ¼ 0.018, I2 ¼ 57.1%). Among patients with PIK3CA mutated tumours, however, aspirin use was associated with a statistically significant 29% risk reduction in overall mortality (SRE ¼ 0.71, 95% confidence interval 0.51e0.99, P ¼ 0.04) and no heterogeneity among studies was seen (Q chi-squared ¼ 3.09, P ¼ 0.5, I2 ¼ 0.0%). Conversely, among patients with the wild-type PI3KCA gene, there was no significant effect of aspirin with respect to overall mortality (SRE ¼ 0.93, 95% confidence interval 0.61e1.40, P ¼ 0.7), but high heterogeneity between studies (Q chi-squared ¼ 14.72, P ¼ 0.002, I2 ¼ 80%). Meta-regression analysis showed no significant interaction of PIK3CA mutational status on aspirin efficacy (P ¼ 0.4). For exploratory purposes, we also calculated SRE for CSS, even though there were only two studies reporting risk estimates for this end point. There was a trend towards a benefit for aspirin in PI3KCA mutated subjects (SRE ¼ 0.37, 95% confidence interval 0.11e1.32, P ¼ 0.1), albeit with high heterogeneity (Q chi-squared ¼ 3.41, P ¼ 0.07, I2 ¼ 70.7%) (Figure 3).

Numerous prospective observational studies [28] and long-term follow-up of randomised trials of aspirin for cardiovascular prevention showed that regular aspirin use is associated with a reduced incidence and mortality of CRC [1,3], as well as diminished risk of distant metastases for CRC [2,29e32]. Despite this robust epidemiological evidence [33], it remains to be proven if aspirin use can influence the prognosis for patients after diagnosis of CRC and therefore be incorporated in the adjuvant treatment of this deadly disease, which has not improved since the introduction of oxaliplatin more than 10 years ago. Moreover, there is uncertainty as to whether aspirin may be useful to all CRC patients or be limited to specific tumour subtypes. As recent data suggested that regular aspirin use improves overall survival and CSS in the subset of CRC patients with PIK3CA mutated [4e8], we carried out the current meta-analysis to assess the survival benefit of aspirin use after CRC diagnosis according to PIK3CA status. Our results suggest that post-diagnosis aspirin use overall is associated with a weak reduction in overall mortality but a significant reduction in all-cause mortality in PIK3CA mutated CRC, supporting the notion that PIK3CA mutation may be a predictive biomarker of response to adjuvant aspirin therapy [34]. This is biologically plausible given the notion that aspirin has been shown to suppress cancer cell growth and induce apoptosis by blocking the PI3K pathway and this effect might therefore differ depending upon the PIK3CA status [21,22].

Please cite this article in press as: Paleari L, et al., PIK3CA Mutation, Aspirin Use after Diagnosis and Survival of Colorectal Cancer. A Systematic Review and Meta-analysis of Epidemiological Studies, Clinical Oncology (2015), http://dx.doi.org/10.1016/j.clon.2015.11.008

L. Paleari et al. / Clinical Oncology xxx (2015) 1e10

7

Fig 2. Forest plot of the association between post-diagnosis aspirin use and colorectal cancer (CRC) mortality in PIK3CA mutated tumours, PIK3CA wild-type tumours and overall effect.

Fig 3. Forest plot of the association between post-diagnosis aspirin use and cancer-specific survival (CSS) in PIK3CA mutated tumours, PIK3CA wild-type tumours and overall effect. Please cite this article in press as: Paleari L, et al., PIK3CA Mutation, Aspirin Use after Diagnosis and Survival of Colorectal Cancer. A Systematic Review and Meta-analysis of Epidemiological Studies, Clinical Oncology (2015), http://dx.doi.org/10.1016/j.clon.2015.11.008

8

L. Paleari et al. / Clinical Oncology xxx (2015) 1e10

The strength of our meta-analysis is the inclusion of all the published studies of post-diagnosis aspirin use and survival of CRC patients in accordance with the variant PIK3CA gene, a context that is similar to the adjuvant setting. However, our study also has some important limitations that warrant further discussion. First, the number of studies involved in the current analysis is still small and the studies were mostly observational studies, which are susceptible to selection, information bias and confounding. Specifically, the reasons why patients received aspirin after CRC diagnosis were poorly defined, although they may have been related to cardiovascular indications. Moreover, the starting time of aspirin use was not clear in all studies, inasmuch as Chae et al. [5] and Kothari et al. [7] identified aspirin users by use at a time point at or soon after diagnosis, which may not reflect overall post-diagnostic aspirin use, whereas Domingo et al. [6], Liao et al. [4] and Reimers et al. [8] also included as aspirin users patients who were not taking aspirin at diagnosis and started the treatment at a later point and this is prone to immortal time bias. Second, information on CSS was lacking in three [5,6,8] of the five studies, which weakens the overall conclusion, because total mortality is less sensitive in terms of antitumour effect and be subject to dilution of any real effect. Furthermore, one study reporting CSS did not provide information regarding the effect of aspirin on wild-type tumours. Third, in the studies included in the meta-analysis, the dose and duration of aspirin varied from 100 mg [6,8] to 350 mg day [4] or not reported (5). In addition, many patients’ related data were not available, resulting in an increased source of heterogeneity. Although the optimal dose and duration of aspirin as adjuvant therapy of CRC is unknown [28,35], pharmacological data suggest that a daily low-dose aspirin (75e100 mg) was sufficient to accomplish an adequate antineoplastic effect [36,37]. Also, the antiplatelet effect, which has been regarded as a main mechanism for the anti-metastatic effect, can be attained at doses of 75e100 mg day [4,27]. Fourth, CRC is a heterogeneous disease with different molecular characteristics that would lead to different responses to therapy and specifically to aspirin [38,39]. Factors such as stage distribution may be a source of heterogeneity. For instance, 277 of 4589 (6.0%) had stage IV CRC in three studies [3,6,7] and this might represent a confounding factor (see Table 1 for tumour characteristics). Specifically, the best response to aspirin treatment in terms of overall survival in mutated cancers (hazard ratio ¼ 0.29; 95% confidence interval 0.04e2.33) was reported by Domingo et al. [6], where stage IV patients were not included. Conversely, in the study of Kothari and colleagues [7], where stage IV represented 24% of the total population, a lower response to aspirin in PI3KCA cancers was noted (overall survival, hazard ratio ¼ 0.95; 95% confidence interval 0.55e1.63) However, this may also be due to the shorter median follow-up compared with the other studies (Table 1). Another issue in the current study is the lack of information on the anatomical site of the CRC, as PI3KCA

mutations have been reported to be associated with specific clinicopathological features and molecular events, such as KRAS mutation and microsatellite instability [40]. It has been described that, compared with the wild-type form, PIK3CA mutated CRCs are predominantly distributed in the proximal colon (54% versus 34%, P > 0.001) [41], with a gradient of mutational frequency that decreases from the cecum to the rectum [41,42]. In the studies included in this meta-analysis, only Kothari et al. [7] noted that left-sided cancers trended towards better overall survival and CSS, although this was not significant. Interestingly, aspirin has been associated with a greater benefit at the proximal colon (hazard ratio ¼ 0.35, 95% confidence interval 0.20e0.63) [1,2]. Additional sociodemographic factors, such as age, may affect the therapeutic efficacy of post-diagnosis aspirin use. Population studies support an increased incidence of most cancers among older adults and CRC has high prevalence in the aging population, with a median age around 70 years at diagnosis and 75 years at death [43]. About 50% of all individuals undergoing CRC surgery relapse and die of metastatic disease [44,45]. Although adjuvant chemotherapy has significantly improved the prognosis of individuals with high-risk colon cancer in subjects 70 years old or younger, the effect of adjuvant chemotherapy in older patients is less clear [46e48]. In the study by Reimers and colleagues [8], the effect of aspirin on overall survival was greater in patients older than 70 years not receiving adjuvant chemotherapy (Adjusted Relative Risk, ARR ¼ 0.59, confidence interval 0.44e0.88). Finally, different loci of PIK3CA mutations might be associated with a different tumour behaviour and response to aspirin. For instance, a significant negative correlation between PIK3CA mutation in exon 20 and the response to anti-epidermal growth factor receptor (EGFR) antibodies has been observed [49]. Likewise, PIK3CA exon 20 mutations were associated with poorer progression-free survival, overall survival and lower objective response to antiEGFR antibodies, whereas patients with exon 9 mutations were equally responsive to wild-type subjects [50]. Although the association between a specific site of PIK3CA mutation and the response to aspirin has not been addressed in any of the studies analysed in the present meta-analysis, future studies should prospectively assess this association. Our findings are heavily influenced by the striking effect noted by Liao et al. [4], which observed an 82% relative reduction of CCS in their cohort of PIK3CA mutated tumours. This finding has no clear explanation but it deserves to be confirmed in randomised controlled trials; one potential reason is that data were derived from two selected health professional populations that do not represent the general population. Although the association between a specific site of PIK3CA mutation and the response to aspirin has not been prospectively addressed in any of the current studies, two adjuvant trials of aspirin are underway or about to be started, the ASCOLT trial [51] and the ADD-ASPIRIN trial [52], to assess the efficacy of aspirin in the adjuvant setting [53]. In the ADD-aspirin trial, a subgroup analysis on

Please cite this article in press as: Paleari L, et al., PIK3CA Mutation, Aspirin Use after Diagnosis and Survival of Colorectal Cancer. A Systematic Review and Meta-analysis of Epidemiological Studies, Clinical Oncology (2015), http://dx.doi.org/10.1016/j.clon.2015.11.008

L. Paleari et al. / Clinical Oncology xxx (2015) 1e10

mutated PIK3CA tumours is planned. However, results will not be available for 5 years or more. Although the overall estimate of an 18% mortality reduction is less than reported by Rothwell et al. [1], the confidence interval is wide and the lower end of the 95% confidence interval gives a 37% reduction, which is compatible with their findings of a 37% mortality reduction. Also, we used overall survival as the main outcome, whereas Rothwell et al. [1] used CSS, which could be a major dilution factor in our study. As we note, only 50% of patients died from CRC in our reviewed population, so the impact on CRC death could be twice as large if there was no effect on other causes. In conclusion, our findings suggest that the benefit of post-diagnosis aspirin treatment on overall mortality in CRC is more pronounced in PIK3CA mutated tumours, although the low number of studies prevents definitive conclusions. Adjuvant trials addressing this issue are warranted.

[11]

[12] [13] [14]

[15]

[16] [17]

Acknowledgements Supported by the Italian Association for Cancer Research (AIRC, grant no. 15468, PI Andrea DeCensi).

[18]

[19]

References [1] Rothwell PM, Wilson M, Elvin CE, et al. Long-term effect of aspirin on colorectal cancer incidence and mortality: 20-year follow-up of five randomized trials. Lancet 2010;376(9754):1741e1750. [2] Rothwell PM, Wilson M, Price JF, et al. Effect of daily aspirin on risk of cancer metastasis: a study of incident cancers during randomised controlled trials. Lancet 2012;379:1591e1601. [3] Rothwell PM, Fowkes FG, Belch JF, et al. Effect of daily aspirin on long-term risk of death due to cancer: analysis of individual patient data from randomised trials. Lancet 2011;377(9759):31e41. [4] Liao X, Lochhead P, Nishihara R, et al. Aspirin use, tumor PIK3CA mutation, and colorectal-cancer survival. N Engl J Med 2012;25(367):1596e1606. [5] Chae YK, Kunwha K, David SH, et al. PIK3CA mutation, aspirin use and mortality in patients with metastatic colorectal cancer participating in early-phase clinical trials. Cancer Res 2013;73(Suppl. 1, abstract 164). [6] Domingo E, Church DN, Sieber O, et al. Evaluation of PIK3CA mutation as a predictor of benefit from nonsteroidal antiinflammatory drug therapy in colorectal cancer. J Clin Oncol 2013;31:4297e4305. [7] Kothari N, Kim R, Jorissen RN, et al. Impact of regular aspirin use on overall and cancer-specific survival in patients with colorectal cancer harboring a PIK3CA mutation. Acta Oncol 2015;54(4):487e492. [8] Reimers MS, Bastiaannet E, Langley RE, et al. Expression of HLA class I antigen, aspirin use, and survival after a diagnosis of colon cancer. JAMA Intern Med 2014;174(5):732e739. [9] Labelle M, Hynes RO. The initial hours of metastasis: the importance of cooperative host-tumor cell interactions during hematogenous dissemination. Cancer Discov 2012;2:1091e1099. [10] Palumbo JS, Talmage KE, Massari JV, et al. Platelets and fibrin(ogen) increase metastatic potential by implementing

[20]

[21]

[22]

[23]

[24]

[25]

[26] [27]

[28]

[29]

[30]

9

natural killer cell-mediated elimination of tumor cells. Blood 2005;105:178e185. Erpenbeck L, Schon MP. Deadly allies: the fatal interplay between platelets and metastasizing cancer cells. Blood 2010;115:3227e3236. Gay LJ, Felding-Habermann B. Contribution of platelets to tumor metastasis. Nat Rev Cancer 2011;11:123e134. Hanahan D, Weinberg RA. Hallmarks of cancer: the next generation. Cell 2011;144(5):646e674. Aleksandrova K, Jenab M, Bueno-de-Mesquita HB, et al. Biomarker patterns of inflammatory and metabolic pathways are associated with risk of colorectal cancer: results from the European Prospective Investigation into Cancer and Nutrition (EPIC). Eur J Epidemiol 2014;29:261e275. Tsilidis KK, Branchini C, Guallar E, et al. C-reactive protein and colorectal cancer risk: a systematic review of prospective studies. Int J cancer 2008;123(5):1133e1140. Wang D, Dubois RN. The role of COX-2 in intestinal inflammation and colorectal cancer. Oncogene 2010;29(6):781e788. Eberhart CE, Coffey RJ, Radhika A, et al. Up-regulation of cyclooxygenase 2 gene expression in human colorectal adenomas and adenocarcinomas. Gastroenterology 1994;107(4):1183e1188. Ogino S, Kirkner GJ, Nosho K, et al. Cyclooxygenase-2 expression is an independent predictor of poor prognosis in colon cancer. Clin Cancer Res 2008;14(24):8221e8227. Vogel LK, Saebo M, Hoyer H, et al. Intestinal PTGS2 mRNA levels, PTGS2 gene polymorphisms, and colorectal carcinogenesis. Plos One 2014;9(8):1e8. Samuels Y, Wang Z, Bardelli A, et al. High frequency of mutations of the PIK3CA gene in human cancers. Science 2004;304(5670):554. Kaur J, Sanyal SN. PI3-kinase/Wnt association mediates COX2/PGE(2) pathway to inhibit apoptosis in early stages of colon carcinogenesis: chemoprevention by diclofenac. Tumour Biol 2010;31(6):623e631. Uddin S, Ahmed M, Hussain A, et al. Cyclooxygenase-2 inhibition inhibits PI3K/AKT kinase activity in epithelial ovarian cancer. Int J Cancer 2010;126(2):382e394. Di Popolo A, Memoli A, Apicella A, et al. IGF-II/IGF-I receptor pathway up-regulates COX-2 mRNA expression and PGE2 synthesis in Caco-2 human colon carcinoma cells. Oncogene 2000;19(48):5517e5524. Stroup DF, Berlin JA, Morton SC. Meta-analysis of observational studies in epidemiology. A proposal for reporting. JAMA 2000;283:2008e2012. Moher D, Liberati A, Tetzlaff J, et al. Preferred reporting items for systematic reviews and meta-analyses: The PRISMA Statement. J Clin Epidemiol 2009;62:1006e1012. DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7(3):177e188. Sterne JA, Sutton AJ, Ioannidis JP, et al. Recommendations for examining and interpreting funnel plot asymmetry in metaanalyses of randomised controlled trials. Br Med J 22 Jul, 2011;343:d4002. Cuzick J, Thorat MA, Bosetti C, et al. Estimates of benefits and harms of prophylactic use of aspirin in the general population. Ann Oncol 2015;26(1):47e57.  C, Rostom A, Lewin G, , et alUS Preventive Services Task Dube Force. The use of aspirin for primary prevention of colorectal cancer. A systematic review prepared for the US Preventive Services Task Force. Ann Intern Med 2007;146(5):365e375. Algra AM, Rothwell PM. Effects of regular aspirin on longterm cancer incidence and metastasis: a systematic

Please cite this article in press as: Paleari L, et al., PIK3CA Mutation, Aspirin Use after Diagnosis and Survival of Colorectal Cancer. A Systematic Review and Meta-analysis of Epidemiological Studies, Clinical Oncology (2015), http://dx.doi.org/10.1016/j.clon.2015.11.008

10

[31]

[32] [33]

[34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

L. Paleari et al. / Clinical Oncology xxx (2015) 1e10 comparison of evidence from observational studies versus randomized trials. Lancet Oncol 2012;13:518e527. Bosetti C, Rosato V, Gallus S, Cuzick J, La Vecchia C. Aspirin and cancer risk: a quantitative review to 2011. Ann Oncol 2012;23(6):1403e1415. Thun MJ, Jacobs EJ, Patrono C. The role of aspirin in cancer prevention. Nat Rev Clin Oncol 2012;9:259e267. Langley RE, Burdett S, Tierney JF, et al. Aspirin and cancer: has aspirin been overlooked as an adjuvant therapy? Br J Cancer 2011;105(8):1107e1113. Langley RE, Rothwell PM. Potential biomarker for aspirin use in colorectal cancer therapy. Nat Rev Clin Oncol 2013;10(1):8e10. Smith T, Elwood P, Keating C, et al. The Aspirin Foundation Scientific Conference: the history, the present state and the future of aspirin prophylaxis. Ecancermedicalscience 2014;8:388. Ruffin MTIV, Krishnan K, Rock CL, et al. Suppression of human colorectal mucosal prostaglandins: determining the lowest effective aspirin dose. J Natl Cancer Inst 1997;89:1152e1160. Sample D, Wargovich M, Fisher SM, et al. A dose-finding study of aspirin for chemoprevention utilizing rectal mucosal prostaglandin E(2) levels as a biomarker. Cancer Epidemiol Biomarkers Prev 2002;11:275e279. Bosman FT. WHO classification of tumours of the digestive system, 4th ed. Lyon: World Health Organization, International Agency for Research on Cancer; 2010. Perez VB, Romera LA, Hernandez PS, et al. Colon cancer molecular subtypes identified by expression profiling and associated to stroma, mucinous type and different clinical behavior. BMC Cancer 2012;12:260. Lia X, Morikawa T, Lochhead P, et al. Prognostic role of PIK3CA mutation in colorectal cancer: cohort study and literature review. Clin Cancer Res 2012;18(8):2257e2268. Yamauchi M, Morikawa T, Kuchiba A, et al. Assessment of colorectal cancer molecularfeatures along bowel subsites challenges the conception of distinct dichotomy of proximal versus distal colorectum. Gut 2012;61:847e854.

[42] Rosty C, Young JP, Walsh MD, et al. PIK3CA activating mutation in colorectal carcinoma: associations with molecular features and survival. Plos One 2013;8(6):1e9. [43] Ferlay J, Steliarova-Foucher E, Lortet-Tieulent J, et al. Cancer incidence and mortality patterns in Europe: estimates for 40 countries in 2012. Eur J Cancer 2013;49(6):1374e1403. [44] McCleary NJ, Dotan E, Browner I. Refining the chemotherapy approach for older patients with colon cancer. J Clin Oncol 2014;32(24):2570e2580. [45] Bastiaannet E, Sampieri K, Dekkers OM, et al. Use of aspirin postdiagnosis improves survival for colon cancer patients. Br J Cancer 2012;106:1564e1570. [46] Andre T, Boni C, Mounedji-Boudiaf L, et al. Oxaliplatin, fluorouracil, and leucovorin as adjuvant treatment for colon cancer. N Engl J Med 2004;350:2343e2351. [47] Jessup JM, Stewart A, Greene FL, et al. Adjuvant chemotherapy for stage III colon cancer: implications of race/ethnicity, age, and differentiation. JAMA 2005;294:2703e2711. [48] Sargent DJ, Goldberg RM, Jacobson SD, et al. A pooled analysis of adjuvant chemotherapy for resected colon cancer in elderly patients. N Engl J Med 2001;345:1091e1097. [49] de Roock W, Claes B, Bernasconi D, et al. Effects of KRAS, BRAF, NRAS, and PIK3CA mutations on the efficacy of cetuximab plus chemotherapy in chemotherapy-refractory metastatic colorectal cancer: a retrospective consortium analysis. Lancet Oncol 2010;11(8):753e762. [50] Yang Z-Y, Wu X-Y, Huang Y-F, et al. Promising biomarkers for predicting the outcomes of patients with KRAS wild-type metastatic colorectal cancer treated with anti-epidermal growth factor receptor monoclonal antibodies: a systematic review with meta-analysis. Int J Cancer 2013;133(8):1914e1925. [51] Ali R, Toh HC, Chia WK. ASCOLT Trial Investigators. The utility of aspirin in Dukes C and high risk Dukes B colorectal cancer e the ASCOLT study: study protocol for a randomized controlled trial. Trials 2011;12:261. [52] http://www.addaspirintrial.org/. [53] Phillips I, Langley R, Gilbert D, et al. Aspirin as a treatment for cancer. Clin Oncol 2013;25(6):333e335.

Please cite this article in press as: Paleari L, et al., PIK3CA Mutation, Aspirin Use after Diagnosis and Survival of Colorectal Cancer. A Systematic Review and Meta-analysis of Epidemiological Studies, Clinical Oncology (2015), http://dx.doi.org/10.1016/j.clon.2015.11.008