Clinicopathological cohort study of kidney biopsy findings resulting in dialysis during long-term follow-up exceeding 30 years

The Japanese Society for Dialysis Therapy publishes an annual report presenting the results of a questionnaire survey conducted at dialysis facilities. A 2018 report identified diabetic nephropathy, chronic glomerulonephritis, and nephrosclerosis as the three primary cause of dialysis in Japan and noted an increasing prevalence of dialysis among older adults [1]. In this report, patients with a history of diabetes mellitus (DM) were classified as having diabetic nephropathy (DN); those with hematuria and/or nephrotic-range proteinuria were classified as having chronic glomerulonephritis; and those with hypertension accompanied by minimal proteinuria and atrophic kidneys were classified as having benign nephrosclerosis (BNS), alternatively termed arteriosclerosis/arteriolosclerosis without glomerulopathy. The United States Renal Data System, which compiles the causes of end-stage kidney disease (ESKD), reported in 2012 that underlying kidney diseases included diabetes (38%) and hypertension-related nephropathy (25%) [2]. However, it is important to note that the results of both the Japanese and US reports were based on the subjective assessments of the attending physicians rather than on kidney biopsy findings.

Studies based on the kidney biopsies have yielded significantly different results from those described above. A retrospective cohort study conducted in the UK, which included 463 patients aged > 70 years who underwent kidney biopsy, reported that vasculitis and paraproteinemia were the most prevalent diseases leading to dialysis, whereas primary glomerulonephritis was the least common [3]. This report excluded diabetic nephropathy, possibly because of the UK-specific criteria for kidney biopsy [3].

There is a paucity of studies utilizing histopathological evaluations of kidney biopsies to determine the most common underlying diseases resulting in dialysis initiation. Consequently, we conducted a retrospective cohort study of patients who underwent kidney biopsy and subsequently progressed to ESKD requiring dialysis initiation. Owing to the rigorous development of guidelines in recent years, the trend was observed by dividing the cohort into three groups according to the time periods and analyzed. Additionally, we analyzed and described the baseline clinical characteristics at the time of kidney biopsy and investigated independent risk factors associated with early dialysis initiation using multivariate Cox regression analysis. Furthermore, we analyzed the clinical characteristics of the two major causes of dialysis initiation, DN and IgAN, and compared the two groups.

This retrospective cohort study elucidated histologically diagnosed kidney diseases responsible for ESKD. Tables 1, 2, and 3 describe the clinical characteristics of the study cohort. Lower eGFR, higher proteinuria, DN, and advanced age were significant risk factors for early dialysis initiation (Table 4). DN and IgAN are the two primary causes of ESKD. We analyzed the clinical characteristics of DN and IgAN and compared these two diseases at the time of kidney biopsy, as illustrated in Figs. 1, 2, and 3.

Although the number of patients on dialysis has been reported to be increasing in many countries, few studies have evaluated the underlying diseases, as confirmed by kidney biopsy, that led to dialysis initiation [1‐3]. In patients with DM, accurate diagnosis of kidney histology is a crucial indicator of the future course of treatment when the kidneys are affected. Diagnosing DN using classification schemes facilitates improved communication between renal pathologists and clinicians, provides a logistical structure for prognostic and interventional studies, and enhances clinical management and efficiency. In accordance with this approach, Tervaert et al. proposed a pathologic classification of DN [6]. Furthermore, the Committee on Practical Guidance for Kidney Biopsy in Japan published a second edition of its kidney biopsy guidebook, suggesting that in patients with DM and renal complications, a kidney biopsy can provide a wide range of information, including: (1) presence of typical DN, (2) existence of arteriolar hyalinosis without typical DN, (3) presence of glomerular lesions other than DN, and (4) a mixture of these lesions [7]. Even in patients with clinical information suggesting typical DM accompanied by nephrotic-range proteinuria, retinopathy, and neuropathy, kidney biopsy provides pathological information that can inform treatment strategies to improve renal prognosis. Additional studies on kidney biopsy in DM patients support this approach [8, 9].

Previous reports have indicated that DN, chronic glomerulonephritis, and hypertension-related nephropathy are the three primary causes of dialysis initiation in Japan [1, 2] based on questionnaire surveys. The current study demonstrated that IgAN, FGS, MGN, and MPGN are types of chronic glomerulonephritis leading to dialysis. The pathogenesis of hypertension-associated nephropathy as an initiating factor in chronic kidney disease remains unclear [10, 11]. Moreover, the data presented herein suggest that BNS is not a major direct cause of ESKD in patients requiring kidney biopsy, a finding that is consistent with the study by Sumida et al., who investigated the renal prognosis of patients diagnosed with BNS by kidney biopsy [12]. In support of this, a meta-analysis examining kidney dysfunction in ten trials with 114,000 person-years in patients with nonmalignant or benign hypertension revealed that patients treated for hypertension did not exhibit a lower risk of renal dysfunction [13]. Furthermore, after 8–12 years of follow-up, intensive blood pressure control in patients with hypertension was not superior to normal blood pressure control with regard to renal function [14]. However, patients in the intensive blood pressure control group with proteinuria exceeding 0.22 g/gCre demonstrated a potential benefit of preserved renal function [14], suggesting that the underlying kidney condition, such as chronic glomerulonephritis or glomerulopathy accompanied by proteinuria, is susceptible to hypertension resulting in declined kidney function. Therefore, a combination of hypertension, chronic glomerulonephritis, and proteinuria may be a significant contributor to progressively declining kidney function, ultimately leading to dialysis initiation. As most kidney biopsies in older adults are associated with hypertension, the involvement of hypertension in the underlying kidney disease should be carefully evaluated. Indeed, a previous article reported that kidney biopsy of hypertensive patients revealed that 16% of the patients had primary nephritis, 19% FGS, 43% MNS, and 22% BNS [15]. This suggests that approximately 80% of hypertensive patients have glomerulopathy or glomerulonephritis, a potential factor for poor kidney prognosis, which are histologically distinctive from BNS.

Conversely, others studies have reported that biopsy-proven BNS was associated with poor renal prognosis, although in this particular study, 71% of participants had proteinuria with a maximum proteinuria of 5.1 g/day [16], suggesting that an underlying glomerulopathy could be present in some patients in addition to BNS. As delineated in the methodology section, BNS should present with intact glomeruli, and arterioles or arteries being the primary site of injury. Therefore, by definition, BNS does not typically manifest as proteinuria. This analysis corroborates our finding that BNS is an infrequent direct cause of ESKD in patients requiring kidney biopsy.

Although FGS in young patients has garnered significant attention regarding genetic abnormalities, its occurrence in clinical practice is infrequent, with majority FGS patients being over 50 years of age. Determining whether FGS is secondary to hypertension, or a primary disease, presents a diagnostic challenge. Similarly, ascertaining whether hypertension is secondary to FGS is equally problematic. It is well-established that FGS can develop due to numerous underlying conditions [17], thus rendering it a heterogeneous diagnosis in terms of etiology. The Columbia classification for FGS [18] provides a useful framework; however, in practice, patients often do not meet the specified criteria, resulting in a diagnosis of “not otherwise specified”. Future research is necessary to address this diagnostic limitation.

Several additional considerations merit further discussion in this study: the underlying kidney diseases in other dialysis patients; a more detailed characterization of IgAN patients, including clinical and histological grades at kidney biopsy; the prevalence of cigarette smoking and hypertension, both of which are factors associated with ESKD; a comparison between this study and the Japan Renal Biopsy Registry (J-RBR); a comparison between this study and previously reported [12] BNS prognoses; the potential association between FGS and BNS; complications related to kidney biopsy under the indication criteria in this study; the consistency of indication criteria for kidney biopsy throughout the study period; and the prevalence of BNS among all kidney biopsies, including patients without ESKD.

The underlying kidney diseases in other dialysis patients were also analyzed as described in the results section. ADPKD was the most prevalent renal disease, because our institution specializes in its treatment. Consequently, the distribution of diseases, excluding ADPKD, may be comparable to that of a general institution.

A more comprehensive characterization of patients with IgAN, including clinical and histological grades at kidney biopsy, was conducted as described in Table 6. Furthermore, as illustrated in Fig. 2, the time required to reach ESKD after biopsy tended to be longer in period 3 than in period 1. Since 2000, tonsillectomy and glucocorticoid pulse therapy have been implemented for IgAN at our institution, resulting in a significant improvement in the prognosis of IgAN. Notably, all 16 patients with creatinine levels exceeding 2.0 mg/dL ultimately required dialysis; however, the interval to dialysis initiation was prolonged. Nevertheless, some patients did not receive this treatment because of their advanced age. Although Fig. 4 demonstrated that patients with biopsies after 2000 obtained statistically better survival than their counterparts, Table 7 described only renal function and proteinuria were independent risk factors for early dialysis initiation. This may be due to the fact that patients with biopsies after 2000 presented with worse renal function (median serum creatinine 2.1 mg/dL vs 1.6 mg/dL) and higher-level proteinuria (median proteinuria 1.8 g/gCre vs 1.3 g/gCre).

The prevalence of cigarette smoking and hypertension, both of which are associated with ESKD, is of significant importance. As demonstrated in the results section, the prevalence of smoking was as high as 35.3% for patients who initiated dialysis. Considering that the prevalence of smoking in Japan is 16.7%, this elevated rate may be a contributing factor to the progression of renal failure. Regarding hypertension, 97% of patients who initiated dialysis were taking antihypertensive medication as demonstrated in the results section. This observation suggests that hypertension is associated with progression to ESKD, irrespective of the underlying disease. This finding indicates that treatment modalities targeting the underlying disease in addition to antihypertensive agents may be necessary.

A comparative analysis between this study and the Japan Renal Biopsy Registry (J-RBR) warrants further consideration. J-RBR compiled data on renal biopsies conducted in Japan. The data revealed that IgAN, DN, FGS, MCD, MGN, and other conditions were prevalent across all age groups, whereas AAV and BNS showed an increased incidence in elderly patients. BNS is observed in a significant proportion of elderly patients who present with low proteinuria and preserved renal function. A recent classification of nephrosclerosis into malignant nephrosclerosis and essential hypertension/arteriosclerosis was implemented for data collection. However, there is a dearth of reports on the renal prognosis for each specific disease entity, which remains an area for future investigation within the J-RBR [19‐22].

Regarding a comparison between this study and previously reported [12] BNS prognoses, in 2016, Sumida et al. conducted a study of cases in which the pathologist's report described nephrosclerosis [12]. In this study, the diagnosis was made without EM in cases with predominantly LM and negative IgG and IgA test results. Subsequently, in the present study, kidney biopsy specimens were re-examined by multiple pathologists. In some instances, where the initial diagnosis of nephrosclerosis was based on a high number of global sclerosing glomeruli and a high degree of atherosclerosis, one or more focal segmental sclerosing glomeruli were newly identified. The diagnosis of FGS was reassessed in cases of partial IgM deposition by IF and foot process effacement by EM, and these cases were excluded from the BNS. Additionally, genetic testing was performed in patients with a family history of renal disease, and those diagnosed with Alport or UMOD were excluded from the BNS. Furthermore, patients with onion skin lesions were reclassified as having MNS or Sclero and excluded from BNS based on clinical features, including hyperreninemia and rapid deterioration of renal function. Because of these iterative procedures, the number of nephrosclerosis cases with a final diagnosis of arteriosclerosis (BNS) alone was reduced from the list of diseases causing dialysis induction to one. Conversely, renal prognosis of patients with BNS and reduced renal function who discontinued hospital visits was not observed. Patients who succumbed to extrarenal lesions during the course of the disease were lost to follow-up.

In addition to the aforementioned discussion concerning FGS, the potential association between FGS and BNS warrants further examination. The 34 cases diagnosed with FGS in this study excluded primary FGS, in which the primary disease, such as a hereditary condition that causes FGS, was evident. Consequently, further classification of FGS is not feasible. However, most cases diagnosed with FGS exhibit multilayered fibroelastosis of the renal arteries and are diagnosed as coexisting arteriosclerosis and arteriolosclerosis, suggesting a significant association between FGS and renal atherosclerosis. Furthermore, many patients with ESKD present with high-grade proteinuria, a characteristic that differs from BNS by definition because glomerulopathy does not occur in BNS. This observation may indicate the susceptibility of BNS patients to progress to secondary FGS.

Complications associated with kidney biopsy under the specified indications can be delineated as follows: A Silverman needle was utilized between 1985 and 1997. Subsequently, an automatic biopsy needle is used. During the former period, renal hemorrhage following kidney biopsy was prevalent; however, only one patient underwent endovascular treatment, and no patient required surgical nephrectomy. In the latter period, the incidence of renal hemorrhage after kidney biopsy decreased; nevertheless, endovascular treatment was performed in two patients. The guidelines established by the Japanese Society of Nephrology [7] stipulate that kidney biopsy in patients at high risk of hemorrhage should be conducted with informed consent within the same institution, and that renal biopsy should be performed in an institution where hemostatic treatment following renal bleeding is feasible [23] and when the patient has been provided with comprehensive explanation and consent. This protocol was approved by our hospital.

Consistent indication criteria for kidney biopsy throughout the study period was a critical consideration. This investigation demonstrated that serum creatinine levels at the time of kidney biopsy in diabetic patients were elevated in period 1 compared to those in period 3 (Fig. 1). Kidney biopsy in patients with diabetes is typically performed upon referral by a diabetologist for renal dysfunction. During period 1, numerous patients with diabetes were referred to a nephrologist for dialysis. However, in period 3, patients with diabetes were referred to nephrologists at an earlier stage of CKD to consider CKD management. Consequently, Cre values at the time of kidney biopsy in patients with diabetes decreased. Three patterns of renal involvement are observed in patients with diabetes: typical DN, nondiabetic renal involvement, and a combination of the two. Kidney biopsies were performed to differentiate between these three patterns [7, 24].

The prevalence of BNS among all kidney biopsies, including those of patients without ESKD at our institution, provides additional perspectives. During the observation period from 1985 to 2018, 6763 kidney biopsies were performed at our institution. Among these, 271 (4.0%) were diagnosed with BNS or nephrosclerosis, with no significant findings other than atherosclerosis. Among these 271 patients, one patient progressed to ESKD. Furthermore, the present analysis was based on the cases that had undergone kidney biopsy. Consequently, renal histology in cases where kidney biopsy could not be performed due to the small kidney size remains unknown. Such cases may include BNS. Sumida et al. investigated the clinical picture of BNS in a large number of cases using the J-RBR. They found that BNS was more prevalent in elderly patients with preserved renal function and low proteinuria levels. In our investigation of dialysis cases without kidney biopsy, we observed that the pathogenesis of dialysis in such patients is often influenced by extrarenal lesions, such as systemic atherosclerotic diseases, including cardiac and hepatic conditions, rather than by a gradual decline in renal function. Moreover, certain diseases can be diagnosed without kidney biopsy in cases where the procedure cannot be performed. These include ADPKD, SLE, AAV, AMY, pyelonephritis and genetic disorders such as Alport. In patients with renal failure complicated by DM, the typical histology of diabetic nephropathy can be inferred in cases of diabetic retinopathy; however, it is challenging to assess cases without retinopathy. Kwon provides a valuable reference in this regard [24]. The results of kidney biopsies in patients with diabetes revealed that diabetic nephropathy was diagnosed in 37% of cases, kidney diseases other than diabetic nephropathy in 54.3%, and complications in 8.5%.

The strength of this cohort study lies in its extensive follow-up period of 35 years, which facilitated the identification of patients who ultimately developed ESKD or who initiated dialysis. Among the diverse histological diagnoses, we were able to ascertain the kidney diseases listed in Table 1 as the causes of ESKD, indicating that these histological diagnoses at the time of kidney biopsy represent a potential risk factor for subsequent ESKD development.

The limitations of this study include its single-center design, which may introduce bias for kidney biopsy, although the indication for kidney biopsy was consistent for all patients regardless of the presence of DM, and a multicenter study is necessary to obtain a comprehensive representation of the situation. In addition, it is noteworthy that the renal tissue of patients with ESKD who presented to the hospital with atrophic kidneys were not examined, as these patients are not typically evaluated by kidney biopsy. Furthermore, it is often challenging to establish a diagnosis from kidney biopsy in patients with numerous sclerotic glomeruli and very few remaining functioning glomeruli, and the validity of the diagnosis of FGS in a few segmental sclerotic glomeruli in the absence of predominant deposition on IF or EM is questionable. Lastly, the cohort did not include patients who did not required dialysis; therefore, the interpretation of the data should take this limitation into account.

In conclusion, given that the criteria for kidney biopsy were consistent for patients with and without DM at our institution, we posit that the findings of our study provide a comprehensive representation of the nephropathy types that contribute to ESKD.