Acute kidney injury caused by tenofovir disoproxil fumarate and diclofenac co-administration
The renal elimination of tenofovir (TFV) may be subject to renal drug−drug interactions that may increase the risk of kidney injury. Case reports indicated that diclofenac might increase TFV-associated nephrotoxicity via a drug−drug interaction, leading to an increased intracellular TFV concentration in proximal tubular cells.
A retrospective analysis of data for all patients from the Frankfurt HIV Cohort (FHC) who had diclofenac prescriptions between January 2008 and June 2012 was carried out.
Among 89 patients with diclofenac use, 61 patients (68.5%) were treated with tenofovir disoproxil fumarate (TDF) and 28 patients (31.5%) were treated with TDF-sparing combination antiretroviral therapy (cART). Thirteen patients (14.6%) developed acute kidney injury (AKI) shortly after initiating diclofenac treatment. AKI occurred exclusively in TDF-treated patients, although all had previously stable renal function. All cases were accompanied by new onset of at least two parameters indicating proximal tubular damage, such as normoglycaemic-glucosuria and hypophosphataemia. TFV-associated nephrotoxicity was demonstrated by renal biopsy in four cases. Additionally, 11.5% of patients on TDF treatment developed new-onset proximal tubular damage, while having a preserved glomerular filtration rate. In contrast, diclofenac did not affect renal function in patients with TDF-sparing cART, as only one case of isolated hypophataemia was observed in these patients. In univariate analysis, risk factors for AKI were TDF-containing cART (P = 0.0076) and pre-existing hypophosphataemia (P = 0.0086).
Drug−drug interaction caused by diclofenac could exacerbate TFV-associated nephrotoxicity. Diclofenac should be used with caution in patients on TDF therapy, especially in those with hypophosphataemia. Our findings need to be confirmed in larger studies.
Acute kidney injury (AKI) is a common, often reversible complication with an estimated incidence of 2.7 to 5.9 per 100 person-years in HIV-infected out-patients [1-4]. Combination antiretroviral therapy (cART) is first-line therapy for HIV-associated nephropathy [3, 5, 6], but some antiretrovirals, including tenofovir (TFV), indinavir and atazanavir, have been associated with acute and chronic renal failure [3, 7, 8].
Renal toxicity related to TFV is caused by proximal tubular injury that may present with various manifestations (hypophosphataemia, normoglycaemic-glucosuria, hypouricaemia, proteinuria and/or metabolic acidosis, including partial or complete Fanconi syndrome) attributable to impaired reabsorption of glomerular filtrate and nephrogenic diabetes insipidus, as well as AKI [9-12]. Tenofovir disoproxil fumarate (TDF) is a prodrug which is rapidly metabolized to TFV in plasma. TFV is eliminated unchanged by glomerular filtration (70–80%) as well as proximal tubular secretion (20–30%). TFV enters the proximal renal tubular cells via human organic anion transporter 1 (hOAT1) and is then actively secreted into the urine by the multidrug-resistance protein (MRP), mainly MRP4 [10, 11, 13, 14].
Nonsteroidal anti-inflammatory drugs (NSAIDs) inhibit drug transport of MRP4 and could increase the risk of TFV toxicity as a result of accumulation in proximal tubular cells when used concomitantly as a consequence of this renal drug−drug interaction (DDI) [10, 11, 13]. In addition, NSAIDs can independently reduce glomerular filtration and thus potentially further contribute to the development of renal injury. In a few cases, acute onset of TFV-associated nephrotoxicity has been reported in patients with previously stable renal function associated with the concomitant use of NSAIDs [15-19]. The purpose of this study was to determine the frequency of and risk factors for acute renal failure and proximal tubular dysfunction after initiation of NSAID therapy in HIV-infected patients receiving cART.
Patients treated from January 2008 to June 2012 in the Frankfurt HIV Cohort (FHC) were evaluated for this study. This time-frame was chosen because since January 2008 all prescription (not just prescription of cART) has been registered electronically. Patients prescribed the oral NSAID diclofenac, with a follow-up period of at least 6 months, were identified and data for these patients were then retrospectively analysed. Clinical data were extracted from the medical records. All serum creatinine values registered prior to prescription of NSAIDs were considered the baseline serum creatinine values, if no significant changes (> 0.20 mg/dL or 17.6 μmol/L) had occurred within the last 12 months. The estimated glomerular filtration rate (eGFR) was obtained using the Modification of Diet in Renal Disease (MDRD) equation . Pre-existing chronic kidney disease (CKD) was defined as being present if the baseline glomerular filtration rate was always < 60 mL/min/1.73 m2 for a period of at least 3 months prior to NSAID prescription, according to current definitions of CKD stage ≥ 3 . AKI and its severity were defined according to AKI Network (AKIN) criteria . Complete renal function recovery was considered to have occurred if the eGFR 12 months after the occurrence of AKI was > 60 mL/min/1.73 m2, or had returned to baseline values in patients with previous CKD . Indicators of tubulopathy, such as hypophosphataemia, glucosuria and proteinuria, were analysed for all patients. Data on serum phosphorus levels were available for all clinical visits and data on urine analysis were available at least every 3 months. Hypophosphataemia was definded as serum phosphorus level < 1.5 mg/dL (= 0.485 mmol/L), glucosuria was defined as present under normoglycaemia, and proteinuria was defined as a protein/creatinine ratio ≥ 200 mg/g (= 22.6 mg/mmol). Data on glucosuria and proteinuria were derived from spot urine tests in all patients.
Chronic hepatitis B virus (HBV) infection was defined on the basis of detectable HBV surface (HBs) antigen and chronic hepatitis C virus (HCV) infection on the basis of HCV antibodies and HCV polymerase chain reaction (PCR). Variables were expressed as median and interquartile range (IQR) or as proportions, as appropriate. Continuous and categorical variables were compared for univariate analysis between groups using the Mann−Whitney U-test and Fisher's exact test, respectively. All P-values reported are two-sided. Statisticalsignificance was assumed when the P-value was <0.05.
Between 1 January 2008 and 30 June 2012, a total of 89 patients were treated with diclofenac. The majority were male (76.4%) and Caucasian (91.0%). Sixty-one patients (68.5%) were treated with TDF-containing cART and 28 patients (31.5%) were treated with TDF-sparing cART. CKD and proteinuria were present in some of the patients treated with TDF-sparing cART, whereas no patient treated with TDF had either of these comorbidities. Twelve patients were receiving concomitant therapy with other potentially nephrotoxic drugs: trimephoprim-sulfamethoxazole dosed for Pneumocystis carinii pneumonia (PCP) prophylaxis (10 patients), mesalazine (one patient) and allopurinol (one patient) (Table 1).
Table 1. Risk factors for acute kidney injury (AKI)
|Age (years) [median (IQR)]||48.0 (45.0–52.0)||48.5 (42.0–56.0)||0.92|
|Male [n (%)]||12 (92.3)||56 (73.7)||0.17|
|Caucasian [n (%)]||13 (100)||68 (89.5)||0.60|
|Body mass index (kg/m2) [median (IQR)]||23.7 (20.0–27.5)||23.9 (21.5–27.3)||0.60|
|Smoker [n (%)]||6 (46.2)||35 (46.1)||0.31|
|Alcohol [n (%)]||3 (23.1)||14 (18.4)||0.71|
|Injecting drug user [n (%)]||4 (30.8)||9 (11.8)||0.09|
|HIV|| || || |
|Duration of HIV infection (years) [median (IQR)]||9.5 (4.3–15.2)||10.8 (3.8–16.1)||0.96|
|AIDS [n (%)]||5 (38.3)||27 (35.5)||1.0|
|CD4 count nadir (cells/μL) [median (IQR)]||94 (37–233)||137 (31–233)||0.84|
|CD4 count prior to NSAID treatment (cells/μL) [median (IQR)]||388 (264–551)||489 (296–666)||0.33|
|PIs [n (%)]||8 (61.5)||46 (60.5)||1.0|
|NRTIs [n (%)]||13 (100)||61 (80.3)||0.11|
|NNRTIs [n (%)]||1 (7.7)||16 (21.1)||0.45|
|INSTIs [n (%)]||1 (7.7)||12 (15.8)||0.68|
|TDF [n (%)]||13 (100)||48 (63.2)||0.0076|
|Duration of TDF treatment (months) [median (IQR)]||38.0 (20.0–54.5)||25.5 (7.0–57.5)||0.45|
|Atazanavir [n (%)]||1 (7.7)||10 (13.2)||1.0|
|Other drugs|| || || |
|Antihypertensives [n (%)]||3 (23.1)||21 (27.6)||1.0|
|Nephrotoxic drugs [n (%)]||3 (23.1)||9 (11.8)||0.37|
|Trimethoprime [n (%)]||2 (15.4)||8 (10.5)||0.64|
|Comorbidities|| || || |
|Chronic hepatitis B [n (%)]||4 (30.8)||10 (13.2)||0.12|
|Chronic hepatitis C [n (%)]||5 (38.5)||13 (17.1)||0.13|
|Malignancies [n (%)]||1 (7.7)||13 (17.1)||0.68|
|Hypertension [n (%)]||4 (30.8)||22 (28.9)||0.74|
|Diabetes [n (%)]||1 (7.7)||5 (6.6)||1.00|
|Coronary heart disease [n (%)]||0 (0)||11 (14.5)||0.36|
|Peripheral vascular disease [n (%)]||1 (7.7)||5 (6.6)||1.0|
|Cerebrovascular disease [n (%)]||0 (0)||3 (3.9)||1.0|
|CKD (eGFR < 60 mL/min) [n (%)]||0 (0)||4 (5.3)||1.0|
|Prior hypophosphataemia [n (%)]||5 (38.5)||6 (7.9)||0.0086|
|Prior glucosuria [n (%)]||0 (0)||0 (0)||1.0|
|Prior proteinuria [n (%)]||0 (0)||4 (5.3)||1.0|
|New onset under diclofenac treatment|| || || |
|Hypophosphataemia [n (%)]||8 (61.5)||8 (10.5)||0.0001|
|Glucosuria [n (%)]||13 (100)||2 (2.3)||<0.0001|
|Proteinuria [n (%)]||11 (84.6)||3 (3.9)||<0.0001|
AKI following diclofenac treatment occurred in 13 patients (14.6%) (Table 1). Nine patients developed AKI stage 1, three developed stage 2 and one developed stage 3; all within 3 months after diclofenac initiation. All patients with AKI concomitantly developed new-onset normoglycaemic-glucosuria (one patient with known type 2b diabetes mellitus, but well-controlled blood glucose levels and no prior glucosuria). Eleven patients developed new-onset proteinuria. Hypophosphataemia was found in all 13 patients at the time of AKI; it was pre-existing in five and developed de novo in eight. All cases of AKI occurred exclusively in TDF-treated patients. In univariate analysis, development of AKI was associated with TDF-containing cART and prior hypophosphataemia, as well as new onset of hypophosphataemia, normoglycaemic-glucosuria and proteinuria (all P < 0.01).
Acute renal failure was only mild and transient in two-thirds of cases and resolved within < 3 months, but required further medical evaluation and intervention in five patients (four with AKI stage ≥ 2 and one with stage 1). Further investigation of 24-h urine samples from these patients revealed aminoaziduria, tubular proteinuria and metabolic acidosis with normal anion gap in arterial blood gas analysis in all cases. Renal biopsy was performed in four patients and signs of acute tubular damage highly suggestive of TFV were found by histology.
Most patients with AKI (84.6%) had been taking TDF for at least 1 year without renal impairment prior to the use of diclofenac. TDF was permanently stopped in six patients (in all four patients with AKI stage ≥ 2 and in two patients with stage 1). Diclofenac was stopped in all patients with AKI. Of the 13 patients with AKI, eGFR returned to baseline values in eight patients (61.5%), with seven of them remaining on TDF therapy. Renal recovery was incomplete in five patients (38.5%), all of whom had CKD stage 3 more than 1 year after initial AKI, despite stopping TDF as well as diclofenac treatment. Hypophosphataemia as well as normoglycaemic-glucosuria resolved in all patients, and proteinuria completely resolved in seven patients. In the remaining four patients, proteinuria declined after stopping diclofenac and TDF (from a median of 682 to 367 mg/g creatinine), but persisted for > 12 months; all had established CKD stage 3.
In addition to AKI, eight patients (9.0%) with a stable eGFR developed hypophosphataemia. Seven patients were on TDF-containing cART (two with concomitant new onset of proteinuria and normoglycaemic-glucosuria, and one with concomitant new onset of proteinuria only). While normoglycaemic-glucosuria and proteinuria resolved in all patients after the end of diclofenac treatment, hypophosphataemia persisted in three patients, all of whom were maintained on TDF. In contrast, only a single patient with TDF-sparing cART developed intermittent isolated hypophosphataemia concurrent with diclofenac treatment.
In this retrospective study, we found an unexpectedly high incidence of AKI (14.6%) shortly after initiation of diclofenac therapy in HIV-positive patients on cART. Cases of AKI occurred exclusively in TDF-treated patients, although all had stable renal function under this regimen for several months or (in > 80% of cases) even years. All cases of AKI were accompanied by new onset of at least two parameters indicating proximal tubular damage, such as normoglycaemic-glucosuria and hypophosphataemia. TFV-associated nephrotoxicity was demonstrated by renal biopsy in four cases of AKI. In contrast, diclofenac did not affect renal function in patients with TDF-sparing cART. In addition to the patients with AKI, another 9.0% of all diclofenac-treated patients developed new-onset laboratory parameters suggesting proximal renal tubolopathy, while having a preserved eGFR, and all but one of these patients were on TDF-containing cART.
As the main clinical presentation of TFV-associated nephrotoxicity is proximal tubular dysfunction with or without renal failure [9, 10], our findings may be explained by a renal DDI between TFV and NSAIDs increasing the nephrotoxic potential of TFV. However, the observed increase in individual markers of tubulopathy may also be attributable to other concomitant diseases, for example hypophosphataemia caused by chronic diarrhoea, kachexia, vitamin D deficiency or malabsorption. In addition, NSAIDs such as diclofenac can independently induce two different forms of AKI; the most common form is caused by a reduction of renal blood flow mediated by reduced prostaglandin synthesis, and the other, very rare form is caused by interstitial nephritis . However, haemodynamically mediated AKI is not associated with proximal tubular dysfunction, as found in our patients. Interstitial nephritis can indeed cause tubulopathy, but its clinical and laboratory presentation is distinct from that of proximal tubular dysfunction. Patients typically present with haematuria and pyuria, white cell casts or eosinophilia and possibly symptoms of allergic reaction. None of these were found in our patients with AKI. Thus, other types of renal disease cannot fully account for the high incidence of AKI found in the TDF-treated patients. The occurrence of AKI was significantly associated with prior hypophosphataemia, which could be TFV-related, as tubular dysfunction may precede a decline in eGFR, resembling subclinical nephrotoxicity . In addition, laboratory findings suggestive of a TFV-related tubular injury without AKI were also found in 11.5% of TDF-treated patients shortly after diclofenac therapy, arguing in favour of TFV-associated nephrotoxicity as the common pathological mechanism. Finally, renal biopsies confirmed TFV toxicity in four cases .
In addition to AKI related to drug-specific effects of TFV and/or NSAIDs, it was shown in previous studies that TDF, indinavir and atazanavir were associated with the development of acute and chronic renal failure [3, 7, 8, 24]. Yet, in our study no patient had been on indinavir therapy and the numbers of patients receiving atazanavir were comparable between groups (Table 1). Although therapy with other known nephrotoxic agents, such as trimephoprim-sulfamethoxazole, was not more frequent in the TDF group, this does not rule out additive nephrotoxic effects. Only two patients experiencing AKI after initiation of diclofenac treatment had been on trimephoprim-sulfamethoxazole therapy, and diagnosis of renal TFV toxicity was confirmed by renal biopsy in one of them. Moreover, no patient with AKI had been on angiotensin conversion enzyme inhibitor or angiotensin-1 receptor blocker therapy, which is known to increase the nephrotoxic potential of diclofenac. It therefore appears unlikely that the numbers of patients with AKI after diclofenac prescription were attributable to other nephrotoxic drug effects than that of TDF.
AKI under cART has also been associated with prior CKD, AIDS, lower CD4 cell count, HCV coinfection and liver disease in HIV-infected out-patients [1-3] and also with malignancies and traditional risk factors, such as older age, African descent, hypertension and diabetes, in hospitalized patients [11, 25, 26]. However, these risk factors did not differ between HIV-infected patients with and without AKI. In addition, traditional risk factors for AKI such as CKD were even more frequent in patients on TDF-sparing cART.
Withdrawal of the causative agent is the main therapeutic option and may avoid irreversible damage [10, 12]. After discontinuation of diclofenac and/or TDF, renal function did not fully recover in one-third of patients, which is consistent with previous findings on TFV-associated nephrotoxicity [9, 12]. Recovery of renal function was dependent on the severity of AKI in our study, underlining that early detection and withdrawal of the causative agent can avoid irreversible damage [9, 10, 17].
Our study has several limitations, mostly resulting from its retrospective design. First, NSAIDs such as diclofenac are freely available over-the-counter drugs and it is very likely that more patients in our cohort had taken NSAIDs than was reported, resulting in an overestimation of the frequency of AKI. Patients who received a prescription for diclofenac were probably those with higher doses and a longer duration of treatment compared with those who had taken NSAIDs occasionally, but precise data on indication, dose, frequency and duration were unavailable in this relatively small study. A control group of patients treated with TDF but not taking NSAIDs may have been more appropriate, but as NSAIDs are available without prescription such a control group could not be identified reliably retrospectively.
Secondly, we did not assess all parameters according to the gold standard to assess for proximal tubular dysfunction, such as aminoaziduria, metabolic acidosis with normal anion gap or the fractional excretion of phosphate and uric acid in all patients, as these are not routinely tested.
Especially in view of the limited number of patients treated with TDF-sparing cART, our findings need to be confirmed in larger studies allowing multivariate analysis of associated risk factors. Moreover, the impact of other NSAIDs should also be studied in order to assess whether the observed effect is limited to diclofenac.
In summary, we found evidence that co-administration of diclofenac and TDF seems to be a possible contributor to TFV-associated nephrotoxicity. DDI via MRP4 could well be the reason for this finding. We believe that NSAIDs should be used with caution in patients on TDF-containing cART, especially in those with pre-existing hypophosphataemia. Although deterioration of renal function was only mild and transient in the majority of patients, the clinical relevance should not be underestimated: acute renal failure is a strong predictor of morbidity and mortality not just in the general population , but also in the HIV-infected population in the cART era [25, 26]. This increased risk persists even after renal recovery [2, 27-29]. Together with antibiotics, NSAIDs are among the most commonly prescribed drugs with nephrotoxic properties in HIV-infected patients [16, 30]. With an aging HIV-infected population, it can be assumed that the use of NSAIDs will increase over time . Recently developed tenofovir alafenamide (GS-7340) is not a substrate of renal organic anion transporters (OATs) and thus is unlikely to accumulate in proximal tubular cells . Future studies will have to determine whether this might also be associated with less renal toxicity.
Conflicts of interest and funding: MB, PK, CS, TL and PG have received research funding, consultancy fees or financial support for travel to conferences from Gilead Sciences, Bristol Myers Squibb, Boehringer Ingelheim Pharma, Abbott, Abbvie, Roche, GlaxoSmithKline, ViiV Healthcare and Merck Sharp Dome. OJ has received consultancy fees from Gilead Sciences. All other authors have no conflict of interest to declare. No funding was obtained for this study.
Authors' contributions: OJ and MB initiated the study, treated patients, collected and analysed the data and wrote the manuscript. MB and KA examined the renal biopsies. All other authors treated the patients and were substantially involved in the collection and monitoring of data.