The exact prognostic value of a kidney biopsy in the setting of AKI has not been studied previously. This study looks at both clinical and renal biopsy prognostic variables in renal recovery from AKI in the setting of a developing country where the etiology and setting of AKI is often different from that seen in the developed world.
The patient population
The transplant and glomerulonephritis cohorts have been included in other research on AKI(10). However, when looking at the pathology of AKI it was essential to exclude these patients since glomerulonephritis cause specific pathological changes in the kidney which might confound the interpretation of the extent of involvement of the acute kidney injury on the renal biopsy. The most common cause of AKI in this cohort was drug or toxin induced acute kidney injury and septic AKI (Table 1). In data from the Indian society of nephrology AKI registry, sepsis (34.7%) and tropical fevers (9.8%) were considered to be the most common etiology of community acquired AKI followed by AKI associated with liver disease (9.1%) (12). Snake bite AKI is also fairly common in these regions. The difference is the etiological diagnosis in this study when compared to others on AKI is likely because only patients who underwent a kidney biopsy were included. The co-morbidity profile (Table 1) is similar to that observed in the ISN registry with hypertension being the most common co-morbidity and diabetes following closely. When compared to data from developed nations, it is noted that AKI in the developing world occurs more frequently in younger patients with fewer co-morbidities as opposed to the developed world where older populations with more number of comorbidities develop AKI(13). The cohort in this research study were individuals who had severe AKI, that is most of them required dialysis and more than 95% of them had stage 3 AKI. This data is in keeping with prospective studies done in India where more than 50% of patients have AKI stage 3(12, 14).
Follow up and outcome measures
Patients were followed up for a median of 80.50 days (Table 1). Data from other studies showed that only about 13-37.3%(14) of patients with AKI remain on follow up with a nephrologist among Medicare patients(15). Follow up needs to be streamlined for better patient outcomes especially in high-risk patients.
Traditionally it is considered that recovery from AKI follows 4 steps, initiation, maintenance, polyuria and restitution. Renal recovery is expected to occur within 3 month, if it doesn’t if is labelled CKD(16). In this study however, although most patients who stop dialysis do so over a mean of 22.48 days, recovery takes much longer with nadir creatinine generally being achieved over a period of 57 days or more. This is in keeping with data from the mayo clinic services, most patients were able to discontinue dialysis before 6 months while the remaining recovered over a period of 12 months. This may be an important take home message from this data, that recovery occurs slowly and can be anticipated even after 3 months.
A majority (55.6%) of patients on dialysis became free of dialysis during the course of their treatment (Table 2). When compared to western data, this number is low in that among 50% of AKD patients who have survived their initial hospitalisation, only 30% generally required continuation of KRT(17). The higher number in this cohort maybe due to the selection bias of patients. Patients who are likely to not recover AKI, were more likely to require kidney biopsies for prognostication or to provide an alternate diagnosis for the kidney injury.
The clinical characteristic that correlated with renal recovery and outcome measures such as independence from KRT (Table 3) is similar to those documented in another study where underlying CKD and presence of co-morbidities corelate poorly for these measures(18). It is interesting to note that labelling the disease as AKD does not portend a poor prognosis for renal recovery. Another interesting finding is that not just the nadir creatinine that correlates with composite measures of renal recovery but also the time to achieve the nadir. This may give us an early clue likelihood of renal recovery even before nadir creatinine is achieved.
The urine sediment
Pre-renal AKI is known to have a bland sediment, while in ATN urine may show hematuria and muddy brown casts. Automated analysers are less sensitive and specific to detect these casts when compared with a nephrologist interpretation of the urine sediment(19). In this cohort, automated analysers were used to assess urine and the difficulty is often differentiating a severe AKI which has caused ATN from a glomerular disease. This may be a contributing factor to the number of biopsies that were done in the acute phase of the illness to rule out a glomerular disorder.
Outcome measures of AKI
When it comes to measuring renal recovery, there is no dearth of outcome measures(20). The myriad different interpretations of renal recovery and measures make comparing data between studies difficult. Some examples of renal recovery measures that are used in various studies are MAKE DC, 30, 60,90 and 1 year events, while others look at partial recovery (defined variably in different studies) vs. complete recovery. Still other studies look at AKI as transient or persistent based on a 48-hour window. To make the results comparable, we looked at all these outcome variables in AKI, MAKE 30 events were very low, as were MAKE 60, and 90 events (Fig. 2). In data from another trial the incidence of MAKE is approximately 30%(21). These may represent a selection bias in that only patients who survived this initial episode of AKI and were fit for biopsy were included in this study. However, it should be remembered that, most of this cohort were dialysis requiring AKI and the number of MAKE events is much lower than expected. One possible reason is the etiology of AKI which is different than the west and the lower co-morbidity burden of this population. Kellum et al suggested a novel method of quantifying recovery based on timing and degree of recovery. This classification showed the pattern of renal recovery and in other data proved to predict long term outcomes(22). In this cohort the Kellum classes had good correlation with other measures of renal recovery such as complete recovery and independence from KRT (Table 3).
Biopsy findings in AKI
The relative frequency of biopsy findings is shown in Fig. 3. In a systematic review of biopsy findings in AKI, A similar pattern to this one was noted by Wen et al(23). As in this study, tubular sloughing was a common finding followed by epithelial cell flattening and simplification. Very little data is available worldwide correlating biopsy findings with clinically significant outcomes for AKI. One such study was a post-mortem analysis of patients who underwent kidney biopsy for acute kidney injury in COVID-19 patients(24). The only histopathological finding that correlated with severity of AKI and recovery was presence of pigmented casts.
This study provides new insight and findings into correlation of biopsy findings with renal recovery and independence from KRT. As can be expected, chronic vascular sclerotic changes on the renal biopsy have significant correlation with renal recovery (Table 4). This is possibly because the pathophysiology of acute kidney injury is closely linked to renal vasoconstriction, endothelial injury and activation of inflammatory pathways(25). It makes sense that poor blood supply with arteriosclerotic vessels would play an important role in renal recovery. Vacuolisation is commonly seen in drug and toxin induced AKI due to ART (anti-retroviral therapy), CNI (calcineurin inhibitors) and contrast media and hence probably has a specific correlation with etiology and renal recovery(26). Interstitial fibrosis and tubular atrophy in several glomerular diseases has been liked to disease progression(27).
This study is a first-of-its-kind cohort study looking at renal biopsy findings and correlating them with renal recovery however, the selection bias of the study, to include only biopsy proven AKI, will limit the generalisability of the data to all AKI.