Evaluation of 10 Urinary Biomarkers for Renal Safety With 5 Nephrotoxicants in Nonhuman Primates

Animals used in these opportunistic exploratory studies were made available for biomarker research due to colony attrition needs or site closures over the span of several years. Study designs were guided by published literature and limited by available animals. All 5 NHP studies were approved by the Merck Institutional Animal Care and Use Committee and conducted in an Association for Assessment and Accreditation of Laboratory Animal Care International–accredited facility in compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and the Animal Welfare Act. The animals were acclimated and randomized into treatment and control groups. The compounds, species, route and frequency of administration, dose levels, vehicles, and collection days are presented in Table 1. In general, the study details were consistent between protocols and were as follows: male and/or female Macaca mulatta (rhesus) or Macaca fascicularis (cynomolgus) were approximately 3 to 4 years old and weighed 2.5 to 5.5 kg. Doses were calculated based on animal body weight, and the last dose was given approximately 24 hours prior to necropsy in studies with daily dosing. Urine and blood samples were collected at the same or similar time points usually twice as pretest (PT) and then at various intervals (often about 3 days apart) during the study with the last collection at the time of necropsy for blood and overnight prior final necropsy (FN) for urine. Urine was collected by placing individual monkeys in metabolism cages overnight for approximately 14 to 16 hours. Animals were fasted prior and during blood and urine collections. Urine was collected on dry ice for cefpirome study or wet ice for all remaining studies without additional stabilizers or proteinase inhibitors. Samples were aliquoted and stored at −70 °C until subsequent biomarker analyses.

A complete necropsy was performed and select tissues including (at a minimum) kidney, urinary bladder, liver, muscle (rectus femoris), and heart were processed for histomorphologic examination. Additional tissues were examined based on known target organ toxicity when applicable. Tissue sections were fixed in 10% neutral-buffered formalin for approximately 24 hours, processed, and embedded in paraffin. Embedded tissues were cut into 4 to 6 µm sections and stained with H&E. Stained tissue sections were examined microscopically, and grades were assigned using a standardized PSTC Nephrotoxicity Working Group histopathological lexicon with a severity score scale of 0 to 5: 0 (no observable pathology), 1 (minimal or very slight), 2 (mild or slight), 3 (moderate), 4 (marked), or 5 (severe).¹⁵

Serum biomarkers, such as BUN and SCr, and the urinary biomarkers, such as creatinine, albumin, total protein, and NAG, were measured on an automated Roche analyzer. The urinary biomarkers such as KIM-1 (single-plex), OPN, NGAL, and RBP4 (custom 3-plex) were measured using Meso Scale Discovery human assays. For cystatin C, a human enzyme-linked immunosorbent assay (ELISA) was obtained from Biovendor, L-FABP high-sensitivity human ELISA was obtained from CMIC Holdings, and the human assay for clusterin from R&D Systems. The determined urine biomarker concentrations were normalized to urinary creatinine (UCr) values. Samples with values that exceeded the upper limit of quantification (ULOQ) were diluted and reanalyzed. For fold changes calculation purposes, values below the lower limit of quantification (LLOQ) were replaced by half of the LLOQ limit and normalized to UCr. For purpose of data comparisons, cutoff of 3-fold increase from the average of control and PT animal measurements for each individual study was chosen for urinary biomarkers. This cutoff was based on our previous experience in rat¹⁰ with multiple biomarkers. 1.5-fold cutoff for increase in BUN and SCr was based on human diagnostic criteria for AKI.¹⁶

Each of the 5 studies were analyzed separately. All biomarkers were log-transformed to reduce the impact of outliers. Urinary biomarkers were normalized to UCr. Mean and standard deviation were estimated for each treatment group at each time point by the intercept and slope of a linear regression of the values above the limit of quantitation (ULOQ) on the quantiles of the normal distribution.¹⁷ Mean and standard deviation were not estimated when ≥80% of the values in a group were below the LLOQ. For treatment group by time point combinations for which the mean and standard deviation were estimated, a 95% CI was calculated based on an estimate of standard deviation which was pooled across all the time points for that treatment. The 95% CI was computed using the t-distribution with degrees of freedom equal to the sum of the sample sizes across the different time points in the treatment group minus the number of time points. Estimates and CIs were back-transformed to the original concentration scale for presentation purposes. All statistical analyses were performed in R version 3.6.0 (R Foundation for Statistical Computing).

Ten emerging urinary biomarkers were evaluated for their performance across 5 NHP studies with acute DIKI and compared to BUN, SCr, and each other. All biomarker assays were previously assessed in our laboratory for cross-reactivity and performance by their ability to detect biomarker changes in samples with known toxicities. Results of the biomarker analyses for each study are summarized in Figure 1 as estimates of mean biomarker concentration normalized to UCr, with a 95% CI for each treatment group and data collection day. Tabulated results of statistical analysis are provided as Supplemental Table 1. Data for individual animals are summarized in Table 2 where biomarker changes are expressed as fold changes calculated from average of all controls and PTs for a given study.

Five animals (2 males and 3 females) received a vehicle (saline) and 5 animals (3 males and 2 females) received a single intravenous (IV) dose of 2.5 mg/kg of cisplatin. Treated animals were necropsied on study day (SD) 8. Physical signs included emesis, liquid feces, body weight loss, and decreased food consumption. Serum biochemical changes consisted of increased BUN and SCr, slight decreases in phosphorus and calcium, and very slight decrease in sodium. Cisplatin treatment resulted in moderate (2 animals) to marked (3 animals) tubular degeneration.

Histomorphological changes in the kidneys consisted of widespread tubular degeneration and necrosis throughout the cortex and the outer stripe of medulla, most prominently in medullary rays. The findings were centered on proximal tubules but also extended to other regions of the nephron including distal tubules and (rarely) epithelial cells of the collecting duct. Affected tubules showed flattening, disorganization, and clumping of epithelial cells accompanied by area of tubular regeneration characterized by epithelial basophilia and karyocytomegaly. Affected tubules often had luminal cell debris and/or granular and hyaline casts and were occasionally accompanied by focal interstitial cell infiltrates (Figure 2A and B). Degeneration and single cell necrosis of collecting duct epithelium with associated hemorrhage were observed in some animals. No histopathological changes were seen in liver, heart, urinary bladder, testes, adrenal, ovary, or skeletal muscle. Individual animal histopathological findings are provided in Supplemental Table 2.

Cisplatin treatment resulted in increases of all urine biomarkers as well as BUN and SCr. In the majority of animals, the first urine sample collection at SD 3 showed biomarker increases and those increases persisted through SD 8 (Figure 1 and Table 2). When comparing treatment-related effects, the 95% CIs for control and the treated group did not overlap for any of the tested biomarkers with exception of NGAL at SD 3 and cystatin C at SD 8. The most robust increases were observed in KIM-1, clusterin, and albumin, where fold changes from PT reached over 100-fold. Osteopontin, RBP4, and NAG were also increased in all measured samples, although NAG with relatively smaller, on average about 8-fold increases. Neutrophil gelatinase-associated lipocalin, cystatin C, and L-FABP did not increase in animal #102 at SD 3, which presented with positive BUN and SCr changes. The most highly impacted animal (#93) presented with very high biomarker changes at SD 3, which subsequently decreased at SD 6 and SD 8, while at the same time, SCr and BUN increased to 11-fold and 28-fold at SD 8, respectively. Urine production in this animal sharply decreased from 200 and 350 mL at PTs to 17 mL at SD 3 and 30 mL at SD 8. Urinary creatinine concentration remained similar between all 5 samplings from this animal.

The cefpirome study conducted in rhesus monkeys had 4 control animals (2 males and 2 females) receiving vehicle (saline) and 5 treated animals. Two females were treated with a low-dose (400 mg/kg/d) and 3 animals (1 female and 2 males) with a high-dose (800 mg/kg/d) cefpirome sulfate for 7 days. Physical signs consisted of liquid stool and emesis, especially in the high-dose animals. Decreases in alkaline phosphatase and γ-glutamyl transferase (GGT) in serum were observed in the majority of treated animals at SD 8. No histopathological changes were seen in liver, heart, or urinary bladder. Focal degeneration of skeletal muscle was observed at the injection site and was attributed to the IV dosing. Histomorphologic findings following the 7-day treatment period consisted primarily of tubular basophilia (very slight to marked) and less frequently tubular necrosis (very slight to slight), tubular casts (very slight to moderate), and cellular infiltration (slight). Overall, the response to the treatment was variable, and the most severely affected animal had widespread tubular basophilia in proximal convoluted tubules in the cortex with sparing of proximal tubular segments in medullary ray (Figure 3A and B). Many of these basophilic proximal convoluted tubular profiles showed mitotic activity, indicating regeneration. Tubules were also lined by degenerating or necrotic cells, sometimes in clumps, obstructing tubules, and associated with mixed cellular infiltration. Slight tubular necrosis was also present, with moderate numbers of tubular casts: loose granular casts in dilated proximal tubules and hyaline casts in the medulla and distal parts of the nephron. The remaining animals with very slight or slight overall damage had very slight or slight tubular basophilia, focally in the cortex, with/without very slight tubular necrosis or very slight hyaline tubular casts. Individual animal histopathological findings are provided in Supplemental Table 2.

Increases in urine volume and decreases in pH and specific gravity were observed in the majority of treated animals at SD 8. Changes in SCr and BUN were observed only in one high-dose female (#15), and all evaluated biomarkers were increased in both measured time points in this animal except for KIM-1 at SD 3 (Table 2). The second animal with tubular degeneration (#56) did not show increases in SCr or BUN; however, all other urinary biomarkers were increased in at least one time point in this animal. Animal #73 did not present with any kidney findings, and this correlated with a lack of all biomarker responses. The cefpirome study was the only study performed in rhesus, and albumin, total protein, KIM-1, cystatin C, NGAL, and L-FABP were not detectable in the majority of control and PT measurements; nevertheless, biomarkers increased to measurable levels with treatment. The NGAL responses were overall the highest, ranging from 40-fold to 500-fold, and occurred even in animals without degenerative changes, while albumin, total protein, clusterin, KIM-1, and OPN increases generally aligned with microscopic findings of tubular degeneration.

Three male control animals received vehicle (62% castor oil, 33% ethanol in sterile water) and 3 treated male animals received cyclosporine (20 mg/kg/d), subcutaneously, daily for 14 days. Cyclosporine vehicle was not well tolerated and resulted histologically in injection site inflammation and necrosis that were accompanied by changes in serum parameters including very slight increases in albumin and albumin/globulin ratio and accumulation of macrophages with vacuolated cytoplasm in the spleen and lymph nodes. These changes were slightly more pronounced in cyclosporine-treated animals, and these animals also showed increases in serum C-reactive protein. All other clinical signs were of the type seen in untreated animals. Treatment with cyclosporine induced minimal to mild, transmural degenerative changes in scattered small-caliber arterioles in the renal cortex of treated animals. These changes ranged from focal to circumferential fibrinoid necrosis with exudation of fibrin into the surrounding interstitium, to replacement of the arteriolar wall and lumen by whorls of macrophages and fibroblasts. In 2 out of 3 animals, a few cortical tubules directly adjacent to these foci exhibited very slight degenerative changes, with minimal formation of hyaline casts. The most affected tubules had morphologic characteristics consistent with thick ascending loop of the distal tubule with lesser involvement of proximal tubules a pattern of tubular injury often observed secondary to renal ischemia¹⁸ (Figure 4A-E). In addition to the kidney changes, the most severely affected animal also exhibited degenerative changes in medium- to small-caliber arterioles in the tunica muscularis of the stomach, small intestine, and large intestine characterized by minimal to moderate, perivascular to transmural inflammation comprised mainly of neutrophils and macrophages with fewer lymphocytes. In the jejunum, these changes were accompanied by marked transmural necrosis of the affected arterioles. Individual animal histopathological findings are provided in Supplemental Table 2.

The cyclosporine study presented with the most subtle degenerative renal changes as compared to the other nephrotoxicant treatments and therefore allows for a more discriminative biomarker performance assessment. While BUN and SCr did not change during the treatment in any animals, the urine biomarkers such as clusterin, KIM-1, OPN, and RBP4 were increased in both animals presenting with slight tubular degeneration at the time of FN and at time points prior to FN (Table 2) The one cyclosporine-treated animal (#220), which had no microscopic evidence of degenerative renal changes, was likewise negative for any consistent increases in urine biomarkers. When considering the treatment groups, the means of the control and treated group were different only for KIM-1 (SD 4-11), clusterin (SD 4-11), and OPN (SD 8-14).

Everninomicin (ever) and gentamicin (gen) were administrated IV in combination at a low dose of 30 (ever) and 10 (gen) mg/kg/d and a high dose 60 (ever) and 10 (gen) mg/kg/d, with saline as a vehicle. There were 20 animals on the study split into groups with equal number of males and females. Eight treated animals and 2 controls were necropsied on SD 8 after 7 doses, and 8 treated animals and 2 controls were killed after 7 days of recovery on SD 15. One animal did not produce enough urine during the collection period and was not included in the analyses. Hypoactivity was noted beginning on SD 7 or later in monkeys dosed with the high dose. Males from both treated groups had slight body weight losses over the course of the study due to slightly lower food consumption compared to the controls. Dose-related increases in alanine aminotransferase, aspartate aminotransferase (AST), and bilirubin were noted with peak increases on SD 8/9. In addition, serum electrolyte imbalance and a decrease in urine pH were observed. Minimal to marked, dose-related tubular degeneration and/or casts were noted in all test article-treated monkeys with a higher incidence and/or severity in the groups of monkeys euthanized on SD 8. Minimal to moderate, dose-related tubular regeneration was present in most test article-dosed monkeys though at a higher incidence and/or severity at the SD 15 recovery necropsy (Figure 5A and B for end of treatment necropsy and Figure 6A and B for recovery necropsy). Degeneration was characterized by proximal tubule epithelial cell swelling, cytoplasmic hyaline droplets, epithelial sloughing, and necrosis. Regeneration was characterized by epithelial cell cytoplasmic basophilia, crowding, and slight increases in mitotic figures, as well as scattered mononuclear cell accumulations in adjacent interstitium and tubular dilatation with cellular cast and proteinaceous material. Microscopic histologic observations outside of kidney included inflammation at the injection site, hemorrhagic necrosis and/or Kupffer cell hypertrophy of the liver, hemorrhage of the heart and/or adrenal glands, ulceration of the stomach and urinary bladder, and erosion of the small intestine and fibrinous inflammation of the lung. Of these, hemorrhage, mucosal ulceration, and lung inflammation are possibly attributable to azotemia. Individual animal histopathological findings are provided in Supplemental Table 2.

Everninomicin/gentamicin treatment elicited the largest and most consistent biomarker responses as compared to the other studies (Figure 1 and Table 2). The peak urine biomarker response was seen at the end of the drug treatment period (SD 8 for the necropsy group, SD 9 for the recovery group) and reached over 100-fold increases for the majority of the biomarkers. The treatment response was similar in both treatment groups (ever/gen 30/10 mg/kg/d and 60/10 mg/kg/d), especially at the later time points. The CIs suggest that there were treatment-related differences between the control and both treatment groups at all study time points for KIM-1, clusterin, albumin, OPN, and NAG. The low-dose group CI overlaps with controls at SD 3 for total protein, cystatin C, NGAL, RBP4, and L-FABP, while BUN and SCr control CIs overlap with both low- and high-dose groups. When considering fold changes from average of controls and PTs, the biomarkers responding with the highest magnitude of changes were KIM-1, clusterin, cystatin C, and albumin followed by L-FABP, NGAL, and OPN, while the lowest magnitude response was observed with NAG, with maximum increases over controls and PTs of approximately 10-fold. The earliest sampling was at SD 3 where BUN and SCr crossed the positive threshold (1.5-fold over the mean of all controls and PTs) in none and 3 out of 15 samples, respectively, while clusterin crossed the positive threshold (3-fold over the mean of all controls and PTs) in 14, OPN in 13, and KIM-1 in 12 out of 15 samples at SD 3. The remaining biomarkers also outperformed BUN and SCr at this early time point.

Everninomicin/gentamicin treatments included a recovery (treatment-free) arm for 10 animals. The urine biomarker responses generally peaked between SD 6 and SD 9 and gradually decreased with time toward necropsy at SD 15. The decline toward recovery differed among biomarkers; most notably KIM-1 and clusterin remained increased in all animals above control levels throughout the recovery arm to SD 15 (Figure 1). Albumin, NGAL, L-FABP, and total protein levels decreased at SD 15 close to control levels in 6 out of 8 samples, and RBP4, cystatin C, and NAG in 5 out of 8 samples. Cystatin C levels showed the most dynamic changes with over 200-fold changes seen at the peak response at SD 6 and with a rapid return to a mean of 2-fold of PT values at SD 15. Dose-related increases in BUN and SCr started at SD 6, peaked at SD 8/9, and decreased to partial or complete recovery by SD 15.

Naproxen sodium was administered orally at a low dose of 200 mg/kg/d or a high dose of 400 mg/kg/d for 14 days with 0.5% methylcellulose in water as vehicle to 7 female and 1 male animal. There were no vehicle dosed control animals. In the high-dose group, 1 female was killed early on SD 11 due to adverse physical signs and 1 female was found dead on SD 14. At both doses, there were test article-related body weight losses, unsatisfactory food consumption, and decreased skin turgor. The main serum biochemical changes consisted of increases in BUN and SCr that were seen in both dose levels as early as the first sample collection on day 3 (Figure 1A). Additional serum biochemical changes consisted of slight decreases in AST, alkaline phosphatase, GGT, and potassium and an increase in triglycerides. The most prominent histomorphologic change in the kidney consisted of tubular dilatation (Figure 7A and B). This change often occurred in a wedge-shaped pattern and was sometimes associated with cellular casts and crystalline material within the lumen, most frequently observed within the deep medulla suggesting obstruction as a component of the renal insult (Figure 7C). Impacted tubules often had focal areas of slight tubular degeneration and were sometimes surrounded by a mixed interstitial inflammatory infiltrate. Very slight papillary edema was present in the majority of the naproxen-treated animals. Other assessed organs (adrenal, heart, skeletal muscle, urinary bladder, liver, ovary, and testis) were without histopathological findings. Individual animal histopathological findings are provided in Supplemental Table 2.

Increases in BUN and SCr ranged from 1.4-fold to 6.4-fold for BUN and 1.5-fold to 8-fold for SCr. The largest fold increase (BUN 6.4-fold and SCr 8-fold) was at SD 9 in animal #67 that was found dead at SD 14. None of the confidence intervals for BUN and SCr overlapped between PT (there were no control animals in this study) and either dose group. Biomarkers such as clusterin, KIM-1, and OPN were also increased in almost all measured samples, generally by 3- to 100-fold, with clusterin increases being the largest, and as the lack of overlap in CIs suggests, these increases were different from PT. The NAG responses were smaller, on average about 4-fold, but there were treatment-related differences between PT and both treatment groups for all time points. Both the low-dose and the high-dose elicited similar magnitude of biomarker changes, and furthermore, biomarker increases were generally of similar magnitude throughout the duration of the study and did not increase or decrease with continued dosing. Cystatin C completely failed to distinguish between PT and treated samples (Figure 1 and Table 2)

Doses and treatment duration for NHP studies were based on very limited published literature as a repeat study (cefpirome, ever/gen) or were adjusted to our needs from published literature (naproxen, cisplatin, cyclosporine). Studies employed opportunistic use of spare animals available due to planned colony attrition or site closures. As a result, each study utilizes very few animals (in contrast to the well-powered definitive biomarker studies published in rats) and therefore has inherent design constraints around the number of dose levels, necropsy time points, and overall statistical power that could be achieved. Even though carefully planned, some of the studies resulted in higher toxicity than expected. Cisplatin and ever/gen presented with severe kidney histopathologies, and the most consistent and the highest biomarker increases. Animals treated with cyclosporine and cefpirome on the other hand presented with the milder and least consistent kidney histopathologies and generally with little to no change in SCr or BUN. The naproxen study was histopathologically in between these extremes with consistent but not very high urinary biomarker increases and very consistent and robust SCr or BUN changes. This spectrum of nephrotoxic tubular injury seen across the 5 studies allows for assessment of performance differences among biomarkers across studies.

Differences in either assay cross-reactivity or in basal expression levels of biomarkers were observed between species. Cystatin C and NGAL control levels in rhesus macaque were mostly below the limit of detection, whereas these biomarkers were generally detectable in cynomolgus macaque. However, the levels of both cystatin C and NGAL in treated rhesus in the cefpirome study were comparable to levels in naproxen-treated cynomolgus animals with similar grade of tubular injury. Several other biomarkers including albumin, KIM-1, L-FABP, and OPN were below the level of detection in PT and control samples in both NHP species (KIM-1 was not detectable in 90% of controls).

All analyses were performed on previously frozen samples that were stored at −70 °C. Studies were performed over several years (2006-2014), and for some biomarkers (clusterin, KIM-1, albumin, NAG, total protein), analyses were performed at the time of the study. For the remaining biomarkers, the analyses were performed after storage using the same lot of assay. Despite the significant time difference in storage, the measurements of control levels for those biomarkers were comparable. Stability of biomarkers in samples stored for longer time periods is often questioned and almost impossible to answer due to the unavailability or changes in assay kits. To answer the question indirectly, it can be noted that after more than a decade in storage (the cefpirome study), all recently tested biomarkers (OPN, L-FABP, KIM-1, NGAL, cystatin C, RBP4) were able to detect increases in treated sample with minimal tubular injury, suggesting minimal to no degradation under these storage conditions.

Liver-type fatty acid-binding protein is expressed in proximal tubules transporting bound fatty acids to peroxisomes or mitochondria. It has been mostly studied in humans and is emerging as a useful biomarker for detection of AKI or chronic kidney disease.^19,20 To our knowledge, this is the first published evaluation of urinary L-FABP in NHP in response to DIKI. Both in cynomolgus and rhesus, using a human-based assay, the control levels remain below detection limits, but L-FABP does increase in certain samples presenting with kidney histopathology. Liver-type fatty acid-binding protein was increased in samples from animals treated with cefpirome but not from animals presenting with very slight tubular damage in the cyclosporine study. Liver-type fatty acid-binding protein is increased in all samples with more severe kidney toxicity; however, sometimes its response is delayed as compared to clusterin, KIM-1, or OPN. While L-FABP response can be quite robust, it is not as consistently elevated as compared to these other biomarkers across all animals treated with naproxen, everninomicin/gentamicin, and cisplatin. Although L-FABP is not a top performer in our sample set, it may prove useful in a biomarker panel approach as more mechanistic understanding of its regulation is gathered.

Cisplatin is perhaps the most commonly used kidney toxicant in animal models. At the dose level of cisplatin selected, treatment resulted in widespread tubular degeneration and necrosis throughout the kidney cortex and the outer stripe of the medulla, which is consistent with reports of others. It had been shown on tissue sections collected 4 days following treatment of cynomolgus monkeys with a single dose of 2.5 mg/kg and lower doses of cisplatin that cisplatin accumulates with the highest concentration in the tubular epithelial cells located in the cortex and the outer stripe of the outer medulla corresponding to the regions of injury.²¹ The same regions of injury are reported in human with the addition of lesions in the collecting ducts.²² Biomarkers of kidney toxicity in cynomolgus monkeys treated with cisplatin were studied previously.^9,23 Uchino et al⁹ reports no degenerative tubular changes at SD 7 following a single dose of 3 mg/kg IV. Due to the very mild nature of the tubular injury, the response of urinary biomarkers in the study by Uchino et al is not very strong, however of interest are very acute biomarker increases most notably for NGAL (at 2 hours at SD 1) and clusterin (at 16 hours at SD 1). Chen et al²³ (single-dose administration of 2.5 mg/kg) report with an extended time course and necropsies at SD 2, 5, 10, and 21 histopathological outcomes comparable to our results, although there was greater variability on an individual animal basis. While there are no reports of significant changes in BUN and SCr in those studies, in our study, SCr was increased above the 1.5-fold threshold for each animal in all measured samples and similarly BUN was increased in most samples. KIM-1 increases reached 100-fold and clusterin up to 800-fold in our study conducted at a dose level of 2.5 mg/kg with necropsy on day 8, compared to less than 10- and 30-fold increases, as reported by Chen et al.²³ These differences may be attributable to more severe tubular degenerative changes in our study, and possibly, in the case of KIM-1, to the use of a more sensitive assay. It is quite interesting that across all animals on all studies the most highly impacted animal (#93) was seen on the cisplatin study. This animal presented with very high urine biomarker changes at SD 3, which did not persist but instead decreased at SD 6 and SD 8, while at the same time SCr and BUN increased. The large drop in urine output is consistent with the sharp rises in SCr and BUN, but the drop in urinary biomarker concentrations such as clusterin, KIM-1, albumin, OPN, and total protein is unexpected. This one case cautions that an apparent reduction in urinary biomarker concentrations at later points are therefore likely not reflective of recovery but perhaps are reflecting loss of kidney reserve and an inability of the remnant tissue to mount a reparative injury response. A drop in these injury response biomarkers when seen concurrently with a rise in SCr and BUN appears to be heralding a very poor prognosis, reflecting a kidney that cannot recover.

Naproxen nonselectively inhibits both cyclooxygenase 1 (COX-1) and COX-2, which are involved in prostaglandin synthesis. Prostaglandins in kidney are involved in regulation of renal blood flow, glomerular filtration rate, and salt and water excretion. The COX-2 is distributed in kidney in a species-specific pattern and is believed to account for significant species differences in renal sensitivity to naproxen treatment.²⁴ In our study with naproxen, histomorphological findings resembled NSAID-associated kidney changes seen in human, even though such injury in people is rare. Very similar histopathological observations were reported previously in NHP, and although urinary biomarkers were not measured, SCr and BUN increases were noted.²⁵ Azotemia often accompanied by proteinuria is reported clinically as well.²⁶ Interestingly, while BUN or SCr were increased in almost all samples at SD 3, their levels did not increase with time despite continued dosing and remained relatively flat, as did generally the measured urine biomarkers as well. This early and substantial BUN and SCr increase without the expected magnitude of increases in biomarkers that would accompany such changes invites the question about potential prerenal causes for the BUN/SCr increases. The BUN/SCr ratio was variable between time points and did not generally differ from PT.²⁷ Urine volume output did not show any consistent trends. It was concluded previously with naproxen in cynomolgus²⁵ that the azotemia seen in their study was likely renal in origin. The increases in biomarkers, namely, clusterin, KIM-1, and OPN, indicate intrarenal injury, but it is possible that the BUN/SCr increases indicate combination of both prerenal and intrarenal injury. Furthermore, the lack of albumin and cystatin C response shows that naproxen injury did not result in a decrement in tubular reabsorption of filtered urinary proteins and that combined with overall higher clusterin increases compared to KIM-1, and consistent increases in OPN may be pointing to localization of injury to distal tubules.

Everninomicin/gentamicin kidney toxicity in NHP was first demonstrated by Davis et al⁶ where he reported results for both compounds separately and in combination. We replicated Davis’s results for the combination treatment with similar findings of degeneration, necrosis, and hyaline droplets involving tubular epithelium throughout the cortex. This toxicity, typical for aminoglycosides, is similar to that described in the human literature.^6,28 The study by Davis et al was used to evaluate gene expression changes and accessible biomarkers were not examined. In combination ever/gen treatment, clusterin, OPN, and HAVrc-1 (KIM-1) RNA increases were predictive of drug-induced toxicity as early as SD 1; however, BUN and SCr remained unchanged until study end at SD 7. Our treatment groups were in combination only; therefore, we cannot separate everninomicin effects from gentamicin, but they are expected to be similar. Urinary biomarker changes have been evaluated in NHP treated with gentamicin alone previously, and albumin, α1-microglobulin, clusterin, and OPN were found to be outperforming BUN and SCr. Interestingly, in the same report, urinary KIM-1 was not increased, while KIM-1 mRNA expression was highly upregulated, which could be explained by the possibility that a human-based assay was used that does not fully cross-react with NHP.⁸ Neutrophil gelatinase-associated lipocalin was found to be increased as early as 2 hours after gentamicin treatment.⁹ Everninomicin/gentamicin elicited the largest fold changes in urinary biomarkers relative to all of the other toxicants, notably, cystatin C, a poor performer in other studies, increased over 100-fold which may be explained by its competition for megalin, a transporter that is used for gentamicin uptake as well.^29,30 The inclusion of a recovery phase in the study design allowed for monitoring biomarker changes during the treatment-free period. While a majority of biomarkers either returned or were decreasing to PT levels, KIM-1 remained elevated for the longest duration, confirming its role as a biomarker of tubular injury and regeneration.³¹

Cyclosporine is a calcineurin inhibitor used in organ transplantation. While there are no previous reports of studies with cyclosporine in NHP, cyclosporine toxicity has been extensively studied in humans where structural changes in the afferent arterioles of the kidney are a prominent feature and occur secondary to vasoconstriction with subsequent tissue ischemia/hypoxia leading to cellular injury.^3,4,32,33 A similar pattern to that described in the human literature was also observed in our NHP study with degenerative and hyaline changes found in small arterioles, most prominent in the deep cortex, with associated perivascular inflammatory infiltrates and tubular degeneration directly adjacent to the affected vessels implying an ischemic etiology. This study had the least severe finding of degeneration of tubules, and only 4 biomarkers, such as clusterin, KIM-1, OPN, and RBP4, were increased in the second week of the dosing when compared to PT and controls. It is provocative that both naproxen and cyclosporine, which may both be adversely impacting the vascular component of the nephron to reduce renal blood flow, present with reduced effects on biomarkers associated with protein and cystatin C reuptake in renal epithelial tissue.

Cefpirome is a cephalosporin antibiotic, which as a group known for potential nephrotoxicity. Cefpirome was initially believed to be safe for kidneys,³⁴ but it was established that in the older patient, administration of cefpirome could cause SCr increases.² Publications on cefpirome effects on kidney are very scarce. Our cefpirome study design was guided by a published study,³⁵ where cefpirome administered at 500 mg/kg/d in rhesus and 800 mg/kg/d in cynomolgus resulted in severe kidney toxicity within a week of dosing in a subset of animals. We observed relatively wide variability in the severity of nephrotoxicity observed in individual animals. Cefpirome was the only study performed in rhesus monkeys, and some differences in detection of control sample were noted especially for cystatin C and NGAL, where almost all control values were below the level of detection. The treated samples, however, increased to similar levels as in the cynomolgus studies. Neutrophil gelatinase-associated lipocalin was the best responding biomarker in this study, but severity of the tubular injury did not correlate well with the magnitude of fold changes. This could be artifact of the fold calculations since the majority of the biomarker measurements were below quantification levels at PT measurement and therefore, for calculation purposes, were replaced by ½ of the LLOQ value and normalized to UCr and therefore may not accurately reflect actual values.

In conclusion, NHP treatment with nephrotoxic drugs resulted in kidney histomorphological changes that were generally similar to those reported in humans. This extensive evaluation in NHP further allowed for comprehensive description of the toxicities in and beyond kidney. Serum chemistry analyses and urine analyses of emerging kidney safety biomarkers were also performed and compared to the histopathology outcomes in each study. All 10 evaluated biomarkers were preselected for their promising performance in rats and/or humans and therefore performed well in the detection of PT kidney damage in NHP. Subtle differences were noted for some biomarkers in their performance in cynomolgus and rhesus, but all biomarkers responded well to toxicity in both species. Clusterin and KIM-1 were the most sensitive in separating controls from treated at early time points and outperforming BUN and SCr. Clusterin and KIM-1 also were also increased in the very subtle multifocal periarteriolar tubular changes induced by cyclosporine injury. The first-time evaluation of L-FABP in NHP demonstrated increases with tubular kidney injury, although L-FABP control levels were undetectable. This extensive evaluation in NHP linking histopathological changes for common and mechanistically diverse nephrotoxic drugs with translational urinary kidney safety biomarkers performance further strengthens the supporting data for using these urinary kidney safety biomarkers as sensitive tools to detect and monitor kidney toxicity in drug development studies in NHP and as translational tools to further ensure patient safety.