Capecitabine is an oral 5-fluorouracil pro-drug fluoropyrimidine chemotherapeutic agent. [1] Oral chemotherapy has been emerging as a more convenient alternative for the conventional IV route drugs without diminishing the possible clinical benefit achieved by the IV agents[2] [38, 39]. In a questionnaire done on around 400 patients who have previously received both oral and IV chemotherapy regimens, a major preference for the oral route was seen in around three-fourth of the patients. This preference is mostly due to the lower alteration of daily life routine, less hospital waiting time, less IV-related complications, and less worry about IV access-related difficulties [38].
One of the most commonly associated adverse events associated with capecitabine is gastrointestinal upset; commonly treated with PPIs or other forms of acid-suppressing drugs [9]. However, PPIs have long been avoided with capecitabine due to some evidence of interference with its pharmacokinetics and efficacy [6, 7]. However, a review recently done by Cheng et al. has reported lack of evidence regarding this notion [13]. The suggested interaction between PPI and capecitabine is mostly due to the pre-notion that the increase in the gastric PH by the PPIs could possibly affect the gastric fragmentation of the tablet, and therefore affect its rate of absorption [8, 12]. However, capecitabine tablets can dissolve over multiple PH degrees ranging from the highly acidic spectrum up to an almost neutral environment [13], and therefore the average gastric pH while on PPI – around 4 – is not sufficient to significantly affect the ionization, and absorption, of capecitabine.
Proton pump inhibitors have long been studied for possible extra-acid-suppression benefits. They have proven to bear anti-inflammatory as well as possible anti- resistance benefits in case of multi-drug resistant cancers [10, 18, 40]. However, one of the recently studied benefits includes a recent study by Hiromoto et al. that has reported significant reduction in the severity of the HFS (p < 0.05), possibly being due to the reduction in the TNF-α in the mice limbs (p < 0.01). This study also had a retrospective patient-based arm that was also included in the meta-analysis data of HFS and has reported significant reduction in the HFS in people that were taking concomitant PPI – with an odds ratio of 0.74 in favour of PPI use [4].
Given the contradicting results regarding the benefits and the risks of using PPIs concomitantly with capecitabine; we have tried to meticulously assess both in our meta-analysis, to account for the already-present discrepancy within the literature.
Qualitative assessment of each included study regarding different safety outcomes revealed significant findings in three out of the 14 studies. However most of these studies, eight of the 14, are retrospective in origin, with another three being secondary analyses of prior trials, making their scientific evidence of lower value when compared to actual primary clinical trials with confounder-control [41].
Another possible drawback of the included retrospective studies is that all of the study data were based only on drug dispensal data, with some studies including patients in the PPI group if they received PPIs at any point during treatment [8, 25], therefore exposing these studies to a form of selection bias. Another discrepancy is seen in the study by Chu et al., which showed significant differences only in the incidence within the capeox-only arm [12]; while the capeox/lapatinib arm showed no difference with the use of PPIs. This might raise questions regarding the validity of such results, given that lapatinib does not cause HFS in the first place[42]. Additionally, in the study by Wong et al. [25], confounder adjustment reversed the statistical significance of the effect of PPIs on the RFS (HR: 2.20; 95% CI: 1.14–4.25; p = 0.18). This is not to in any way suspect the validity or credibility of the studies’ methodologies or significant findings, but to point out the common possible limitations, like all of the data obtained from retrospective studies [41, 43].
To this end, the data from our meta-analysis concerning the efficacy has shown that the concomitant use of PPIs was associated with a decline in the OS (HR 1.12; p = 0.05); PFS (HR 1.14; p = 0.008); and RFS (HR 1.75; p = 0.003). Yet, the significance in the PFS effect; both the unadjusted and adjusted hazard ratios, was abolished on using the random effect analysis to account for sample heterogeneity – I2 at 61% and 79%, respectively. The RFS was the parameter scoring the highest HR in response to concomitant PPI administration (HR = 1.75; p = 0.003), that was even higher when adjusted for confounding factors; to reach 1.89; p = 0.005; Supplementary File 2 Fig. 9. Disease-free survival reported no significant differences between both groups; Fig. 4d; Supplementary File 2 Figs. 5, 10.
When it comes to safety prognosis, PPIs were associated with lower incidence of HFS; RR 0.77; p < 0.00001; Fig. 5. These findings were in line with the findings of the study by Hiromoto et al. that contributed to the majority of the weight of this analysis at a sample size as large as 60,668 patients with a RR of 0.75 [4].
Finally, when we analysed the pharmacokinetic overlap between both drugs, no significant correlation was found. The lack of significant difference could in our opinion be attributed to the lower number of studies, as well as the different units and times each study measured the PK parameters after the start of therapy, causing high variations in the levels between the studies (Supplementary File 2 Figs. 13–20). On one hand, the study by Sekido et al. measured the plasma levels on the first day of the first cycle [29]; while the study by Roberto et al. measured them at week 4 and week 8 [17], and the study by van Doorn et al. measured them on day 8 of each phase [30]. This could have actually created major discrepancies in the pharmacokinetic comparison across studies.
Despite the possible detrimental effects of PPIs on survival, our findings concerning the incidence of HFS are in our opinion quite promising; even if not directly. The results of our analysis might open the doors for future studies to fully discover and make use of the exact mechanism by which PPIs reduce HFS. Therefore, could this open the doors for the use of anti-TNF agents in patients taking capecitabine? Particularly given the fact that multiple studies have reported the lack of cancer development or progression in patients diagnosed with IBD – with even potential benefits in osseous metastases as well as overcoming treatment resistance to multiple agents [44–47].
Possible limitations in our meta-analysis include different follow-up durations in the studies, and notable discrepancy in the pharmacokinetic data, including the time of assessment since beginning treatment; as well as the measurement units, which required extensive and meticulous conversions as well as using subgroups within each study in order to fully analyse the already-scarce data in the three assessed studies.
Individual data incriminating the use of PPIs with capecitabine is quite limited – with possible confounders and validity threats in multiple studies seen during our qualitative assessment, due to study design issues, as previously mentioned. However, given the present fear of conducting a clinical trial in case of a positive association, this only leaves us with the possibility of doing a meta-analysis in order to get a better insight on these contradictory findings, as done in this paper.
In conclusion, our meta-analysis on this large population – reaching as much as 3303 patients in the survival analyses and reaching 62,173 patients in the HFS incidence assessment – has reported both beneficial as well as detrimental interactions with capecitabine. Proton pump inhibitors are associated with lower incidence of HFS compared to the control group; with 33% relative risk reduction in the incidence of HFS. However, regarding possible survival risk, PPIs have shown statistically significant worse treatment outcomes in all aspects, save for the DFS, with a much greater impact on the RFS in non-metastatic cases with up to 75% higher relative risk of recurrence at a HR of 1.75; p = 0.003; and increasing to up to 87% increased risk when adjusted for confounders. This should in turn warrant caution and awareness on the possible risks of concurrent use of PPIs along with capecitabine, with extra-caution and meticulous history taking in patients taking capecitabine in the (neo)adjuvant setting due to the much higher impact on patient survival.