PPIs and CKD association present after exclusion of AKI

PPIs have been associated with AKI including acute interstitial nephritis; CKD development; and CKD progression to ESRD. That is, in over 10,000 Atherosclerosis Risk in Communities participants followed for at least 10 years, incident CKD developed among PPI users with a hazard ratio of 1.5 compared to non-users in models that adjusted for co-morbidities, demographics, and concomitant medications. In addition, the current researchers have published similar results in their earlier work, showing that new PPI users are at a higher risk for poor renal outcomes even against a 1:1 propensity score-matched cohort of non-users and users of H2 blockers. This was covered in detail by Praveen Malavade on RFN last year and the number 2 story on Top Nephrology Stories of 2016.

It was previously thought that AKI events were largely driving the association seen between PPI use and CKD.

A recent study from Xie et al in Kidney International aimed to examine the relationship between PPI use and renal outcomes over five years in those without AKI.

Methods: The researchers gathered a cohort of 144,032 veterans (125,000+ new to PPIs and 18,000+ new to H2 blockers). They then created survival models with Kaplan Meier curves to show survival probability for their four renal outcomes: incident eGFR < 60 ml/min/m2, incident CKD, eGFR decrease >30%, and ESRD or >50% decrease in eGFR. Participants were censored at the time of AKI. All efforts were made to exclude subjects with AKI, even before cohort entry. Of note, AKI was defined in four different ways, but the researchers were not able to detect unrecognized AKI (i.e., outpatient AKI that occurs and resolves between lab checks). Covariates used in their multivariate logistic regression model included numerous co-morbidities, medication use such as NSAIDs or ACEI/ARBs, and patient characteristics such as age, race, and BMI.

Results:
New PPI users had an increased risk of

• all renal outcomes compared to users of H2 blockers including eGFR < 60 ml/min/m² with a hazard ratio of 1.19 (95% CI 1.15-1.24)
• incident CKD with a HR of 1.26 (95% CI 1.20-1.33)
• eGFR decline >30% with a HR of 1.22 (95% CI 1.16-1.28)
• ESRD or >50% decrease in eGFR with a HR of 1.30 (95% CI 1.15-1.48).

Discussion: The researchers conclude their data suggest a true association between PPI use and CKD even in those patients that do not develop an AKI along the way. Besides AKI, other possible, underdeveloped explanations for the observed relationship between CKD and PPIs include altered gut microbiome; reduced cell regeneration; upregulated heme oxygenase-1; and increased oxidative stress. The researchers conclude that further investigation is needed and that caution should be applied when considering long term PPI use as well as monitoring kidney function in PPI users.

Personal reflection: Observational data from two cohorts (ARIC participants and veterans) analyzed by two research groups (Lazarus et al and Xie et al) show an association between PPIs and CKD. While this work is not based on randomized controlled trial data nor has it been replicated numerous times over by several different investigators, it is noteworthy. Yet, association is not causality. Should clinical practice change? And to what degree should it change given the many other considerations that can influence renal function including diabetes and hypertension? While we wait for more evidence, my practice has been to discuss PPI use with my patients and encourage alternatives when possible if no clear indication for the medication exists. This is stems from the above information as well as an effort to reduce pill burden and improve medication compliance. Specifically, I ask patients to discuss PPI use with their primary care doctors (or whichever provider started the medication) and suggest the following: weaning down and then off of the medication; avoiding known food triggers; and using H2 blockers as needed.

Melissa Makar, Nephrology Fellow, Duke

Proton Pump Inhibitors: Indigestion for nephrologists!

Proton pump inhibitors (PPIs) are one of the most commonly prescribed medications, mostly because of their efficacy and their favorable safety profile. However, it has been estimated that between 25% and 70% of these prescriptions do not have an appropriate indication. The duration of PPI use also frequently extends beyond recommended guidelines. More than 15 million Americans used prescription PPIs in 2013, costing more than $10 billion. Specifically, in patients with kidney disease, a small study suggested a high consumption of AST (acid suppressing therapy, mostly PPIs) in patients on dialysis (compared to other chronic diseases) and the use of AST had, in the majority of cases, no adequate indication. DOPPS data showed a relatively consistent overall prescription rate of 36% to 38%, suggesting that prescription rates in this population are several times greater than in the general population. It is speculated that the overuse in patients with CKD may be related to polypharmacy, co-morbidities, hypergastrinaemia or the involvement of many physicians in their care.

In 1992, a sentinel case-report identified the PPI omeprazole, as a possible cause of acute interstitial nephritis (AIN) in an elderly woman. Since that time, a growing body of literature suggests that this class of drugs may be linked to AKI, which can potentially lead to chronic injury or kidney failure.

A large nested case–control study supported the association between renal disease and PPI exposure and found that patients with an incident renal disease diagnosis were nearly twice as likely (OR 1.72, p < 0.001) to have been exposed to PPIs, providing the first estimate of an effect size for this relationship.

A subsequent Canadian population-based cohort study, involving nearly 600,000 patients, found that those who commenced treatment with PPIs had a more than twofold increase in rates of AKI (13.49 v. 5.46 per 1000 person-years; HR 2.52) and AIN (0.32 versus 0.11 per 1000 person-years; HR 3.00) in hospital admission diagnoses. Similar results were found in an analysis stratified by individual PPIs, substantiating the class effect. Indeed, more than half (59%) of the patients in this study, on discharge from the hospital, received another prescription for a PPI in the ensuing 100 days. Of these patients, 7.5% were re-admitted to hospital with AKI in the ensu­ing 120 days, consistent with a causal role of PPIs in some instances, providing the first estimate of the risk of drug re-challenge. This also highlights the lack of awareness among clinicians of the potential association between these drugs and renal dis­ease.

In a recently published article in JAMA looking at PPI use and the risk of CKD in two large cohorts, (n=10,482 and n=248,751) PPI use was associated with incident CKD in analysis adjusted for demographic, socioeconomic, and clinical variables (HR, 1.50); and in analysis with PPI use modeled as a time-varying variable (adjusted HR 1.35). The association persisted when baseline PPI users were compared directly with H2 receptor antagonist users (adjusted HR 1.39) and with propensity score–matched non-users (HR 1.76). Twice-daily PPI dosing was associated with a higher risk than once-daily dosing, suggesting a biological gradient. Again, twice-daily PPI dosing was associated with a higher risk of AKI than once-daily dosing.

Of interest to the renal community, PPIs also cause hypomagnesaemia, the cause which is probably related to magnesium wasting via TRPM-6 and TRPM-7 inhibition. This phenomenon is also seen in dialysis patients.
One may argue that all of these are observational studies which can only show an association and hence cannot prove causality (between PPI use and AKI/CKD or hypomagnesaemia). Many of the studies used hospital diagnosis code, which are less sensitive, and may underestimate the true incidence and association. In addition, they rely on administrative data and hence had no access to laboratory indices of clinical or medication data that may have influenced the risk of renal outcomes, including over-the-counter NSAID use (which might perhaps be one of the most common reasons for PPI use).

In conclusion, observational cohort studies represent one of the best methods to study adverse effects of medications used in real-world settings. Given the millions of individuals who take PPIs each year and the fact that more than half of such prescriptions may not be clinically indicated, even a small absolute increase in risk of AKI/CKD outweighs any benefits that might be derived from these drugs in many patients for whom they are prescribed. Although it should not deter clinicians from prescrib­ing PPIs for patients with well-defined indications and durations, the current evidence underscores the importance of ongoing efforts to curtail the indiscriminate use of these drugs.

It also leaves us with some unanswered questions:
1. What is the mechanism of CKD due to PPI use? Is it due to repeated AKI, a hitherto unreported ‘chronic interstitial nephritis’, or related to hypomagnesaemia?
2. Will treatment of hypomagnesaemia change hard outcomes like mortality/ hospital admissions? An interventional study in hemodialysis patients that reported a beneficial effect of oral Mg administration on decreasing carotid intima-media thickness and a recent small randomized pilot study showed that supplementation with Mg carbonate for 1 year in HD patients led to improved calcification scores in some patients. If yes, to what targets?

Post by Praveen Malavade

PPI-Associated Hypomagnesemia

At renal grand rounds today, we heard about a case of severe hypomagnesemia in a patient who was on chronic omeprazole therapy. While the etiology of the low Mg2+ was likely multifactorial, I was surprised to find out that the PPI could have played a causal role. To date, there have been several case series of patients with profound hypomagnesemia that appeared to be due to PPI treatment. The most recent came from Hoorn et al, in which four patients with PPI-induced hypomagnesemia were described. All four had been on PPI therapy for at least one year, and serum Mg2+ ranged from 0.16 to 0.68 mEq/L before repletion. Hypomagnesemia resolved as early as two weeks after discontinuation of the PPI in one case, and took up to three months in another patient. Concomitant hypokalemia was also present in all four cases, and normalized with correction of the serum magnesium.

The mechanism of PPI-induced hypomagnesemia is unclear, although it is hypothesized that somehow the drugs interfere with gastrointestinal absorption. However, data from one case report suggest that a renal effect may also contribute. Regolisti et al described a patient with hypomagnesemia while on pantoprazole. An intravenous infusion of magnesium was used to determine the kidney’s maximum tubular magnesium reabsorption threshold, or the serum ultrafilterable magnesium concentration below which the kidneys retain most of the filtered magnesium. The serum and urine Mg were measured during a period of IV Mg infusion, and the point at which the urine Mg began to increase was deemed the point at which the renal tubular mechanisms for Mg reabsorption were overwhelmed. The maximum tubular reabsorption threshold for Mg was markedly less (0.90 mEq/L) in the patient compared to normal persons (≥ 1.20 mEq/L).

Although the incidence of PPI-induced refractory hypomagnesemia is assuredly quite low, given the infrequent case reports and the large number of people taking PPIs, it should be considered on the differential diagnosis of hypoMg (see Nate’s post) in any patient on a PPI whose low magnesium level is proving hard to correct.

The PPI Paradox

Millions of people take long-term proton pump inhibitors for acid-related GI complaints. These agents are very effective and appear to have an excellent safety profile. However, they may also have interesting effects on calcium and phosphate balance that I recently became aware of.

PPIs act via inhibition of the parietal cell H+/K+ ATPase, producing an ‘inhibitory complex’ which blocks HCl secretion and results in loss of the normal acidic environment of the upper GI tract. Ordinarily, this acidic milieu facilitates the release of ionized calcium from insoluble calcium salts, improving calcium solubility, which is necessary for its absorption. Reduced absorption of calcium is believed to explain the increased rates of osteopenia and fracture seen in medical states associated with achlorhydria, such as pernicious anemia and post-gastrectomy. Furthermore, several studies show that long-term PPI therapy, particularly at high doses, is associated with an increased risk of hip fracture. Also, long-term PPI use can lead to hypomagnesemia with a low fractional excretion of Mg, suggesting a similar mechanism may be responsible.

Based on these observations, I began to wonder if PPIs would decrease the efficacy of calcium-based phosphate binders, via a reduction in the ionized fraction available to bind phosphate in the gut. There is almost nothing written on this, apart from a single, small, prospective Japanese study that did indeed show that acid suppression significantly reduced the phosphorus binding ability of calcium carbonate, resulting in impaired phosphate control in hemodialysis patients.

Then, just to keep things interesting, I found another study that suggests that PPIs may have the opposite effect when given with non-calcium containing binders. In a retrospective analysis of 108 HD patients, more patients taking a PPI plus a phosphate binder achieved phosphorus control, defined as an average phosphorus level below 6 mg/dL and "a calcium × phosphate product level less than 55, than did those taking only a phosphate binder (76% vs 24%; P = .001). However, the benefits were only observed in those taking sevelamer; these authors did not replicate the findings from the aforementioned study from Japan. The mechanism behind this observation are not known.

So there you have it: there is a suggestion from this rudimentary evidence that PPIs may potentially impair phosphate control when given with calcium-containing binders, but paradoxically improve control when given with non-calcium containing binders. This could form the kernel of an interesting clinical study…

The urine's the thing...

Vomiting or nasogastric tube (NG) decompression can lead to metabolic alkalosis, often associated with hypokalemia. When asked what the source of the K loss is, most people assume it is lost in the gastric fluid. However, gastric fluid only contains about 9 mEq/L of potassium, hardly enough to lead to profound hypokalemia. While it is true that cellular shift due to alkalosis could explain some of the hypokalemia, the primary source of potassium loss is via the urine.

Metabolic alkalosis induced by GI loss leads to volume depletion. In this setting, secondary hyperaldosteronism ensues, leading to sodium retention and potassium wasting, hence the hypokalemia. Further, such GI losses are also associated with chloride depletion. Maintenance of electroneutrality usually obligates chloride reabsorption along with sodium retention. But in chloride depleted states, this is not possible. Instead the lumen-negative gradient (due to sodium reabsorption without chloride following) obligates cation excretion, usually potassium or hydrogen ions. Additionally, the hyperaldosteronism increases H+ excretion via effects on the H-ATPase. Together these explain the paradoxical aciduria associated with GI loss-induced metabolic alkalosis as hydrogen is excreted despite alkalemia. The chloride depleted state, as well as the sodium retention induced by volume depletion, also lead to maintenance of the alkalosis by limiting bicarbonate excretion by a variety of mechanisms. It is for this reason that such alkaloses are termed saline responsive, indicated by a low urine chloride - administration of saline leads to correction of hypovolemia and therefore removes the stimulus to aldosterone, while at the same time chloride replenishment allows for the excretion of bicarbonate while minimizing H+ excretion, leading to an appropriately alkaline urine and rapid correction of the electrolyte abnormalities.

But is metabolic alkalosis the only acid-base disturbance that results from NG decompression? Some anecdotal experience suggests not. As noted above gastric fluid contains little potassium and almost no bicarbonate. So at first glance the answer seems to be no. However, there is a not-so-infrequent clinical scenario where NG decompression can result in metabolic acidosis. More distal fluids, such as bile, pancreatic secretions, and small bowel fluids all have high concentrations of bicarbonate (45, 92, and 50 mEq/L respectively). When NG suction is employed to decompress a small bowel obstruction, more distal fluids can be suctioned, leading to bicarbonate loss. In general though, the suctioned fluid will also contain gastric acid and there will be limited net change in acid-base status. However, if gastric acid production is limited by a proton pump inhibitor, metabolic acidosis can ensue. This was the case with a patient we were consulted on with a metabolic acidosis and bicarbonate of 13 in the setting of NG decompression for a partial small bowel obstruction. The patient was also on a PPI. He was not having diarrhea. To test our theory, we checked the pH of his NG fluid (pH testing was more readily available than electrolytes). The pH was 6.9, far above what would be expected from a PPI alone. We suggested holding his PPI. His acidosis resolved quickly, though in the context of discontinuation of NG decompression and bicarbonate administration, so we can't say for sure that holding the PPI during NG decompression solved the problem. Nonetheless, sharing our case with other colleagues revealed similar anecdotal experiences, suggesting that the phenomenon of metabolic acidosis resulting from upper GI loss of bicarbonate in the setting of partial or complete small bowel obstruction, NG decompression, and concomitant PPI administration should at least be kept in the back of the mind.