Discussion
In this prospective population-based cohort study of over 0.47 million participants, we found that regular use of PPI was associated with a 24% increased risk of CKD adjusting for several potential confounding variables. The association was likely to be more marked in men than women. Additional analysis showed that H2RA, a less potent acid-suppressor, was not associated with CKD. Similar findings were demonstrated for the outcome of AKI. The association between PPI use and CKD persists across several sensitivity analyses and did not show clear evidence of variance among the major types of PPIs.
There are increasing evidence from observational studies suggesting that long-term use of PPIs is associated with risk of kidney disease, such as interstitial nephritis, acute kidney injury, and chronic kidney disease (CKD)15,17,24.Lazarus and colleagues using data of 10 482 participants in the Atherosclerosis Risk in Communities study found that PPI users had a 50% higher risk of CKD (HR, 1.50, 95%CI 1.14 -1.96) and validated this findings in a second large cohort of248 751 patients25. In addition, Arora et al18 and Xie et al15 also found that the risk of kidney disease was 10 -28% higher in patients who used PPIs than in non-usersor H2RAs use. Some systematic review and meta-analysis summarized aforementioned three studies and concluded that PPI use was significantly associated with increased the risks of CKD17,24,26. However, lack of adjustment for several important confounders, such as diet, alcohol intake and physical activity, might introduce bias of the findings27. Furthermore, PPIs users might have poor health status, which may were considered as key confounder factors,28,29 but not been controlled in most previous observational studies. In our study, we controlled for these confounders and confirmed that PPI use could increase the risk of CKD.
In addition to aforementioned observational studies, a large randomized controlled trial was conducted to evaluate the safety of PPI pantoprazole among 17 598 participants with a median follow-up of 3 years. The researchers found that pantoprazole was likely to have a modest, although not statistically significant, higher risk of CKD compared with placebo (OR 1.17; 95% CI 0.94 to 1.15).19 However, the trail had also some limitations, such as insufficient power, a short follow-up time frame, potential selection bias and possible conflicts ofinterests20,21. The estimated effect of this trial was smaller than our estimates (HR 1.24). Possible explanations included different population characteristics and follow-up time, the presence of residual confounders, and other biases.
Our results also indicated that women were more likely to a higher risk for CKD associated with PPI use. Similar result was reported that PPI-associated AKI was more frequently among women than men (55.42%vs 44.58%)30. Anothercohort study including over 120 000 individuals indicated that the increased of AKI with statin use was more marked in women than men31. A similar phenomenon was observed that PPI induced hypomagnesemia occurs more commonly among women than men32. Since previous studies have shown that hypomagnesemia was associated with incidence and mortality of kidney disease18,33,34, this phenomenon might explain this gender disparity. Another possible explanation for this gender disparity was that drug-induced nephrotoxicity was associated with immune hypersensitivity reaction35,36, which occurs more commonly in women than men.
The underlying mechanism of the association between PPI use and CKD remain unclear. One possible explanation is that PPIs could increase the risk of acute interstitial nephritis (AIN)37,38, which may transition to chronic interstitial nephritis, and subsequently result in the development of CKD39. Previous study revealed that a significant proportion of patients that suffered PPI-induced AIN did not recover to baseline, having either partial or no renal recovery, possibly because of rapid progression AIN from inflammatory interstitial cellular infiltrates to interstitial fibrosis and chronic interstitial nephritis, especially in those patients with delayed diagnosis or treatment15,39. Therefore, CKD might be a long-term complication and consequence of PPI-induced AIN due to incomplete recovery of renal function and chronic interstitial nephritis39. In addition, PPI use has been linked to hypomagnesemia, which could cause endothelial dysfunction by promoting atherosclerosis, inducing inflammation and inhibiting endothelial proliferation33,40, consequently result in the development of CKD33. In addition, previous studies found that PPI use was associated with the gut microbiota alterations41, and diabetes42, which in turn may increase the risk of CKD33,43. More research is required to investigate the underlying mechanisms.
A notable strength of our study is that this study was based on a well-established prospective cohort, which collected detailed information on lifestyle factors, medications use, and health conditions. We were able to fully investigate potential confounding factors that were often not available in administrative medical databases. In addition, the large sample size and event number enabled us to get precisely estimated effects for individual class of PPIs and subgroups. Lastly, a wide of robust sensitivity analyses enhanced the validity of our findings.
This study had several limitations. First, as an observational study, we cannot fully exclude potential confounding effect. PPI users might have poor health status and be more concerned about their personal health, so they may be more likely to receive CKD related tests. However, the residual confounding effect might be small as our analyses were based on model with comprehensive adjustment for clinical indication PPI use, comorbidities and overall health status. Second, because information about PPI use was collected only once in the baseline, we were unable to investigate the effects of time varying covariates and exposures on CKD. Third, we could not investigate the possible dose-response relationship due to insufficient information on dose and duration of PPI use. Lastly, despite the large sample size, this study may not be able to examine the effect of specific types of PPIs such as esomeprazole, rabeprazole and pantoprazole on CKD due to the relative low number of cases.
In conclusion, this large cohort study found that regular use of PPIs was associated with increased risk of CKD. Although the causal relationship cannot be determined with an observational study, given that the large number of PPI users as well as the potential risk of CKD and other adverse effects such as enteric infections, healthcare providers should carefully weigh up the potential benefits against risk in prescribing PPIs, particularly for long-term continuous use. Further researches are needed to confirm our findings and to explore the possible mechanisms.