Morphological and clinical findings in Sri Lankan patients with chronic kidney disease of unknown cause (CKDu): Similarities and differences with Mesoamerican Nephropathy

In Sri Lanka, an endemic of chronic kidney disease of unknown origin (CKDu) is affecting rural communities. The endemic has similarities with Mesoamerican Nephropathy (MeN) in Central America, however it has not yet been clarified if the endemics are related diagnostic entities. We designed this study of kidney biopsies from patients with CKDu in Sri Lanka to compare with MeN morphology. Eleven patients with CKDu were recruited at the General Hospital, Polonnaruwa, using similar inclusion and exclusion criteria as our previous MeN studies. Inclusion criteria were 20–65 years of age and plasma creatinine 100–220 μmol/L. Exclusion criteria were diabetes mellitus, uncontrolled hypertension and albuminuria >1g/24h. Kidney biopsies, blood and urine samples were collected, and participants answered a questionnaire. Included participants were between 27–61 years of age and had a mean eGFR of 38±14 ml/min/1.73m2. Main findings in the biopsies were chronic glomerular and tubulointerstitial damage with glomerulosclerosis (8–75%), glomerular hypertrophy and mild to moderate tubulointerstitial changes. The morphology was more heterogeneous and interstitial inflammation and vascular changes were more common compared to our previous studies of MeN. In two patients the biopsies showed morphological signs of acute pyelonephritis but urine cultures were negative. Electrolyte disturbances with low levels of serum sodium, potassium, and/or magnesium were common. In the urine, only four patients displayed albuminuria, but many patients exhibited elevated α-1-microglobulin and magnesium levels. This is the first study reporting detailed biochemical and clinical data together with renal morphology, including electron microscopy, from Sri Lankan patients with CKDu. Our data show that there are many similarities in the biochemical and morphological profile of the CKDu endemics in Central America and Sri Lanka, supporting a common etiology. However, there are differences, such as a more mixed morphology, more interstitial inflammation and vascular changes in Sri Lankan patients.


Introduction
There are many similar features of the CKDu endemics in Central America and Sri Lanka, but no proper comparative study of the renal morphology and clinical characteristics has, as of yet, been made. To compare available studies from the two regions is complicated due to differences in study design, semi-quantification of morphology findings, etc. Thus, to make a detailed comparison of the endemics in Sri Lanka and Central America possible, we designed this study of renal biopsies, biochemical and clinical characteristics from patients with CKDu in Sri Lanka using similar inclusion/exclusion criteria and study design as in our previous studies of MeN in Central America [6,7].

Study procedure overview
The study was performed during one week in May/June at Polonnaruwa General Hospital. Blood pressure was tested the day before and on the day of the biopsy. At the morning of the kidney biopsy, blood and urine tests were collected from fasting participants (no intravenous fluids). Ultrasound was performed before and during the kidney biopsy. Participants were interviewed and a questionnaire was filled in. All participants stayed for a 24-hour observation after the biopsy procedure. Within one week, collected samples were transported to Karolinska University Hospital (KUH) for analysis.

Kidney biopsies
Ultrasound-guided percutaneous kidney biopsies were performed using a spring-loaded biopsy needle (16 gauge, Bard Magnum, Bard Biopsy Systems, USA). The biopsies were divided and placed in: 4% buffered formaldehyde, Zeus fixative solution, and 2% glutaraldehyde + 1% paraformaldehyde in 0.1 M phosphate buffer, respectively. Within one week the biopsies were transported to KUH. At KUH, the formaldehyde fixed tissue was embedded in paraffin, sectioned and stained with Ladewig, periodic acid Schiff (PAS), hematoxylin-eosin (HE), and periodic acid silver methenamine (PASM). The tissue in Zeus solution was rinsed and snap frozen, and cryosections were incubated with antibodies against immunoglobulins (IgG, IgA, and IgM), C1q, C3, light chains kappa and lambda, and fibrinogen. Sections from the paraffin-embedded tissue and cryosections were evaluated with polarized light to detect any birefringent crystals. The tissue for transmission electron microscopy (2% glutaraldehyde) was embedded in epoxy resin and then cut into 60 nm thin sections. Histological evaluation with light microscopy, immunofluorescence, and electron microscopy was performed separately by two senior consultants in renal pathology (AW and MS) who thereafter discussed the results to reach consensus. Interstitial fibrosis, interstitial inflammation, and tubular atrophy in the cortical area were semi-quantified similar to Banff criteria [28], as follows: mild, affecting 6%-25%; moderate, affecting 26%-50%; and severe, affecting >50%. Glomerular hypertrophy and vascular pathology were semi-quantified as follows: no/normal, mild, moderate, or severe. When grading podocyte foot-process effacement, the number of slits per micron of glomerular basement membrane (GBM) in five random capillaries per glomerulus was calculated.
Effacement was defined as widespread if !80% of the measured areas had <1.0 slit/μm GBM and segmental if 21-79% of the evaluated areas had <1.0 slit/μm GBM.

Blood samples
The day before the kidney biopsy procedure, blood samples were analyzed for: RBC, WBC, platelets, INR, PT/aPTT, and/or bleeding time/clotting time at the laboratory of General Hospital Polonnaruwa. Just before the biopsy, new blood samples were collected, centrifuged, and the supernatant was transferred to transport vials stored at -40˚C until transport on dry ice to KUH. At Karolinska University Hospital laboratory (KUHL), the samples were analyzed according to standard protocols for glucose, CRP, creatinine (IDMS traceable), BUN, sodium, potassium, magnesium, calcium, phosphate, albumin, alanine aminotransferase, uric acid, cystatin C (IFCC traceable), beta-2-microglobulin, aldosterone, renin, complement (C1q, C3, C3d, C4), antinuclear antibody, and anti-neutrophil cytoplasmic antibody screening, anti-glomerular basement membrane antibodies, and screening for hepatitis B and C and HIV. eGFR was calculated using the CKD-EPI equation for creatinine [29] and creatinine+cystatin C [30].

Questionnaires
A questionnaire was completed by interview of the participants (by CJ). The questionnaire contained open and closed-ended questions concerning medical history (current medicines, use of analgesics, chronic illnesses, snake bites, family history of CKD), work history (current and past work tasks, type of crops cultivated, days working in field per year), work-related chemical exposure (pesticides, herbicides, fertilizers), liquid intake, water source, and smoking habits.

Results
Thirteen patients were recruited to the study. Two patients were excluded because they did not fit the inclusion criteria, one patient due to high fasting blood glucose (16.7 mmol/L) and one patient due to more than one blood pressure medication. In total 11 patients with a mean age of 48 ± 11 years and a mean eGFR of 38 ± 14 ml/min/1.73m 2 were included in the study.

Clinical data and questionnaire
Clinical and questionnaire data are presented in Table 1. All included participants were or had been rice paddy farmers, sometimes in combination with tobacco and vegetable farming. Two Smoking years 11 ± 12 (0-30)

Main source of water last five years
Own well 6 (55%) Village well 2 (18%) Tube well 2 (18%) Main supply 1 (9%) Note: Results are presented as mean ± SD (range) or cases (percentage). a Two patients had increased systolic blood pressure at the morning of the biopsy (145/90 and 160/80 respectively), but were normal the day before the biopsy (130/90 and 130/80 respectively). https://doi.org/10.1371/journal.pone.0193056.t001 participants also worked outside of their own field (one agricultural worker and one pesticide applicator). The mean liquid intake was 4.3 ± 0.9 L/day; the liquid mostly consumed was water (74-97% of the intake), followed by tea (3-25%). Own well was the most common water source. Seven patients stored their water in plastic tanks, two patients in clay pots and two patients in aluminum pots. Medical history and medicine usage are presented in Table 2. No patient used nonsteroidal anti-inflammatory drugs or aspirin. Only three patients used acetaminophen regularly, 2.5 g and 2 g per week and 1 g every other week respectively. Seven participants reported prior use of Ayurvedic medicines, with a mean duration of 0.5 (range 0.08-2) years. All patients reported chemical exposure to pesticides and fertilizers (S1 Table).

Blood test results
Most blood test results are presented in Table 3. CRP was <3 mg/L in all but one patient (Patient 8, 44 mg/L). Four patients had slightly increased complement C4 (0.33, 0.37, 0.43 and 0.46 g/L), which could indicate inflammation, remaining complement levels were normal. No anti-GBM or anti-neutrophil cytoplasmic antibodies were found. Immunofluorescence was positive speckled for antinuclear antibodies in three patients, subsequent analysis for doublestranded DNA, centromere antibodies, and extractable nuclear antigens were negative. Hepatitis B and C, and HIV screening were all negative. Hantaan virus serology was positive for IgG in eight patients and negative for IgM in all patients. Puumala virus was negative for IgG and IgM in all patients. Leptospirosis IgM (ELISA) were positive in four patients, but subsequent analysis with microscopic agglutination test were negative.

Urine test results
Urine test results are presented in Table 4. Urine dipstick showed 1+ albumin in two patients. In sediment, one patient had 8-10 erythrocytes per high-power field, all others had 5. All patients had <5 white blood cells per high-power field. Patient 1 had a few granular casts, no casts were detected in the other patients. No crystals were found. In the two patients with hypokalemia, urinary potassium was >20 mmol/L and FEK was 25% and 23% respectively, indicating renal potassium wasting [39,40]. All patients had FEMg >4% indicating urinary magnesium losses [41]. Urine cadmium, mercury and lead levels were all non-toxic [33][34][35][42][43]. Mean total arsenic was 42 μg/g creatinine, which is higher than in the US [44] but similar to findings in Korea [45], and all samples were below toxic levels (<100 μg/L) [32].

Kidney biopsy morphology
Kidney biopsies were collected from all 11 patients. No adverse events were reported. The specimens contained 7-29 glomeruli (mean 17 glomeruli/specimen). Polarized light on paraffin-embedded tissues detected no crystals. Immunofluorescence showed no signs of immune complex disease. A summary of the light microscopy and electron microscopy findings are presented in Table 5. Light microscopy. Glomerular pathology: Global glomerulosclerosis was found in all patients, mean 43% (range 8-75%) of all glomeruli (Fig 1A and 1D). All participants had glomerular hypertrophy, in most cases mild to moderate (Fig 1B and 1D). No segmental scleroses  or endocapillary cell proliferation was found. Mild mesangial matrix increase was seen in all but two patients, and mild cell proliferation was found in one patient. In seven patients there were signs of glomerular ischemia with thickening of Bowman's capsule and/or wrinkling of the GBM (Fig 1C and 1D), and in three of these cases there was a concurrent moderate intimal thickening in renal arteries (Fig 2B), indicating that the ischemic signs might be caused by vascular pathology. In the remaining four cases the vascular damage was only mild (Fig 2A). Cystic dilatation of Bowman's capsule was seen in three patients (Patients 5, 7, and 8) (Fig 3A). Tubular and interstitial pathology: Tubular atrophy was in most cases mild (Fig 3B). Tubulitis was seen in three patients (Patients 3, 10, and 11), and clusters of intratubular neutrophil granulocytes were found in Patients 10 and 11 (Fig 3D). Interstitial inflammation was of    varying degree-two patients (Patients 1 and 9) had no inflammation, five patients had mild (Fig 3B), two patients (Patients 7 and 11) had moderate (Fig 3A and 3D), and two patients (Patients 3 and 10) had severe inflammation (Fig 3C). The inflammation consisted of mainly lymphocytes. Occasional eosinophils were found in Patient 3. Interstitial fibrosis was mostly mild to moderate (Fig 3A), but one patient (Patient 3) displayed severe interstitial fibrosis. The mild interstitial fibrosis was in all cases focal and the moderate interstitial fibrosis was in most cases diffuse. Vascular pathology: Most specimens had no intimal thickening (Fig 2A), but five had mild to moderate thickening (Fig 2B). Mild smooth muscle hyperplasia was present in four biopsies. Mild to moderate arteriolar hyalinosis was found in all but one biopsy.
Electron microscopy. No immune complex deposits were seen. Two patients showed segmental podocytic foot process effacement (Fig 4A). The majority of the patients had an increased amount of podocytic cytoplasm inclusions such as vacuoles and/or lipofuscin-like bodies (Fig 4B). Bowman's space was normal in all patients. Some of the tubular cells showed reduced brush borders and reduced size.
Individual morphology results. In six patients (Patients 1, 2, 4, 6, 7, and 9) the main finding was glomerulosclerosis, glomerular hypertrophy, and mild to moderate tubulointerstitial changes. Two of the patients also had signs of glomerular ischemia (Patients 6 and 7).
In two patients (Patients 3 and 8) the main finding was chronic interstitial nephritis, with mild to moderate tubular atrophy, tubulitis (Patient 3), moderate to severe interstitial fibrosis, mild to severe interstitial inflammation, and moderate vascular changes.
Pyelonephritis was the main diagnosis in two patients (Patients 10 and 11), with moderate to severe interstitial inflammation, granulocytes in the tubular lumina, tubulitis, mild to moderate interstitial fibrosis, and mild vascular changes. In one patient (Patient 8), the morphology indicated nephrosclerosis with moderate intimal thickening and mild smooth muscle hyperplasia of arteries, signs of glomerular ischemia, and mild to moderate tubulointerstitial changes.

Discussion
In this study we have examined kidney biopsies and biochemical characteristics of 11 male farmers with CKDu in Sri Lanka. We have used similar inclusion and exclusion criteria and study design as in our previous studies of CKDu in Central America, thus enabling a detailed comparison of the two endemics.
The renal morphology in this study showed a more mixed pattern compared to our studies in Central America, where the morphology was relatively homogenous in all cases [6,7]. Six  cases displayed a MeN-like morphology with varying degree of glomerulosclerosis (29-54%), glomerular hypertrophy, mild to moderate tubulointerstitial changes, and mild vascular changes. However, signs of glomerular ischemia, a common finding in our previous cohorts, were only found in two of the six MeN-like cases. Further, no segmental sclerosis was found in this cohort, a finding that we have seen in a few patients in previous studies.
Of the remaining cases, two showed findings indicating active and chronic pyelonephritis with tubulitis, granulocytes found in the tubular lumina, and moderate to severe interstitial inflammation; two showed a morphological pattern of chronic tubulointerstitial nephritis with  moderate to severe interstitial fibrosis and interstitial inflammation; and one showed nephrosclerosis with vascular changes, moderate intimal thickening, mild smooth muscle hypertrophy, and arteriolar hyalinosis.
From our results one can speculate whether these heterogeneous findings in the kidney biopsies represents different stages or severities of one disease or represent different diseases, with multiple etiologies. Nanayakkara et al described in their study of kidney biopsies from CKDu patients in Sri Lanka that the tubulointerstitial damage sometimes was accompanied by massive interstitial inflammation, and they concluded that this was the primary lesion [20]. In another study from Sri Lanka, acute and chronic tubulointerstitial lesions and glomerular scarring were found in patients with acute symptoms (backache, fatigue, joint pain, or dysuria) and kidney injury, indicating that there in some cases may be an acute disease prior to the development of CKD [22]. Although the patients in our study did not present with any acute symptoms, we saw severe interstitial mononuclear inflammation in two biopsies. We also considered if this could be a primary lesion and the other biopsies were later phases. Further studies are needed to elucidate this question. However, the two biopsies with severe inflammation also had chronic tubulointerstitial and vascular changes, indicating a more longstanding disease. One of the biopsies with severe interstitial inflammation was diagnosed morphologically as active and chronic pyelonephritis, but urine cultures were negative. One could speculate whether there could be an unknown pathogen, and we propose that infectious diseases as a cause behind CKDu should be further studied. In Central America, a recently published study from 11 sugarcane workers, with an acute inflammatory illness and kidney injury, reports findings of acute tubulointerstitial disease in kidney biopsies [46]. The authors suggests that the findings are early lesions in Mesoamerican Nephropathy/CKDu in Central America, which also supports that inflammatory and infectious etiologies need to be further investigated.
Due to the findings of interstitial inflammation and signs of pyelonephritis, we analyzed serum for Hantavirus and leptospirosis, common infections in rural Sri Lanka. Notably, elevated Hantavirus IgG-antibody levels, indicating a past infection, were found in 8 of 11 patients. IgM-antibody levels for Hantavirus and leptospirosis were negative. A recent publication from Sri Lanka showed that 55% of patients with CKDu were seropositive (IgG) for Hantavirus compared to 14% in controls but had no prior history of Hantavirus-associated illness [26]. Also in our study, none of the patients reported any hospitalization due to Hantavirusassociated symptoms, suggesting a subclinical infection. Whether a subclinical infection could result in renal scarring or a susceptibility to renal damage is to our knowledge not known; however, studies have reported a higher prevalence of antibodies against Hantavirus in other CKD populations compared to controls [47,48]. We cannot rule out any prior leptospirosis infection in the present study since IgG levels were not measured. Concomitant infections with Hantavirus and leptospirosis has been reported to be common in Sri Lanka [49] and this calls for further studies in the CKDu population. Hantavirus and leptospirosis have been proposed as possible contributing factors in MeN [50], but data on Hantavirus infection in Central America are scarce [51] and no published data have, as of yet, supported Hantavirus or leptospirosis involvement in MeN.
Signs of chronic glomerular ischemia with wrinkling of the glomerular capillaries and/or thickening of Bowman´s capsule were found in seven of the included patients. Chronic glomerular ischemia is often found in ageing and hypertensive kidney disease where intimal thickening of renal arteries are apparent [52,53]. In our study, three patient had moderate intimal thickening in arteries that might explain the glomerular ischemia, but the remaining four had only mild or no intimal thickening, suggesting that other etiologies are responsible for these glomerular lesions. Similar findings with glomerular ischemia not caused by arterial pathology was found in nearly all patients in our previous studies in Central America [6,7].
The kidney morphology in the present study does not resemble Aristolochic acid nephropathy, an endemic nephropathy described in the Balkans [54]. Aristolochic acid nephropathy typically shows a widespread tubulointerstitial disease but normal glomeruli and the disease is correlated to an increased risk of urothelial cancer [55], something that has not been reported in CKDu patients.
In MeN studies from Central America, low serum sodium, potassium and magnesium levels have been common features [6,7]. In the present study, this was also found but less pronounced. However, some patients in this cohort were taking medications that could influence electrolyte levels ( Table 2). In the two patients with hypokalemia, urine potassium levels indicated renal potassium wasting, similar to what we found in MeN patients [7]. Low serum magnesium was common in this cohort, and interestingly, the FEMg was elevated in all patients, indicating renal magnesium losses. However, in CKD patients, FEMg may be increased to prevent hypermagnesemia [56], but this compensatory mechanism unlikely explains the high FEMg found in the patients with hypomagnesemia in the present study. Hypermagnesuria and hyperkaluria have also been reported as a frequent finding in patients with MeN [57], which supports that the diseases are similar. We suggest that the low serum levels of electrolytes are secondary to renal wasting of electrolytes, probably caused by chronic tubular damage. Further, high renal output of electrolytes will make patients more susceptible to salt depletion and/or dehydration due to sweating in hot climates, which could aggravate already-existing kidney damage. A more in-depth analysis of the urine composition should be studied in larger cohorts.
Heat stress with repeated volume and salt depletion has been proposed to be a main factor behind MeN in Central America [58,59], a disease that often affects sugarcane workers. However, while the sugarcane workers in Central America experience a high level of heat stress working in the plantations without shade during a long harvest period, the participants in this study were self-employed farmers working in the rice fields approximately 2-3 weeks per halfyear and had access to shading trees during breaks. Even though the occupational heat exposures differ, both regions have a hot climate and the possibility for heat stress as a contributing factor in CKDu in Sri Lanka should be further studied.
Uric acid has been suggested to be of significance in the pathogenesis of MeN in Central America, since uric acid crystals have been found in the urine of some sugarcane workers during harvest [59,60]. In our study we did not find any crystals in the urine and urine levels of uric acid were normal. However the samples in this study were not collected during work days making a comparison difficult.
Heavy metal exposure has been suggested as a possible etiology behind the CKDu endemic in Sri Lanka. In this study, urine measurements of six heavy metals all showed non-toxic levels, thus not supporting a direct nephrotoxic effect of heavy metals. However, one must take into consideration that urinary heavy metal levels mostly mirrors recent exposure, except for urinary cadmium, which has been found to correlate with the accumulation of cadmium in the kidney [34,61].
In summary, we found that the morphology and biochemical characteristics of CKDu patients in Sri Lanka have many resemblances with the CKDu epidemic in Central America, indicating a similar diagnostic entity. The search for common etiologies affecting both regions should be further studied. However, the mixed morphology pattern seen in Sri Lanka calls for larger biopsy studies with a focus on different possible etiologies affecting renal function in Sri Lanka, such as possible renal-damaging infections. Regarding future studies, we believe that studies of gene and protein expression in biopsy material would give important knowledge about possible etiologies and pathophysiological pathways in the CKDu endemics affecting both Sri Lanka and Central America.
Supporting information S1 Table. Questionnaire results of self-reported chemical exposure to pesticides and fertilizers. N = 11. (DOCX) S2