Clinicopathologic Appearance of Advanced Ketoacidosis With Basal Vacuolation in Renal Tubules: An Investigation for Forensic Autopsy and Retrospective Analysis Open Access

We reviewed the archives of all medicolegal autopsy patients from our department between August 2016 and July 2019. Of these, we evaluated 664 cases for which all organs, including the brain, visceral organs, and serum, could be sampled. Patients' demographic, circumstantial, and clinical information was retrieved from the records of police examinations and family members' contributions or from the primary physician if a patient had a history of visiting a physician. All procedures involving human participants were performed in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This study was approved by the Ethics Committee, University of Toyama, Toyama, Japan (I2020006).

The methods used for the pathologic examination of the heart and central nervous system specimens at our department have been described previously.^11,12  Normal heart weight was calculated from body height and weight.¹³  For the other organs, including the kidney and liver, one or more blocks were sampled per organ. The specimens of all patients were fixed in 20% buffered formalin and routinely embedded in paraffin. Then, 4-μm-thick sections were cut and stained with hematoxylin and eosin, elastica-Masson, or periodic acid–Schiff or underwent IHC. Furthermore, 6-μm-thick sections were cut and stained with Luxol fast blue/hematoxylin and eosin. In addition, if non–paraffin-embedded samples were available, 10-μm-thick sections were cut and stained with Sudan III.

In the kidney, the presence or absence of BV was evaluated using hematoxylin and eosin– (Figure 1, A), elastica-Masson– (Figure 1, C), or periodic acid–Schiff (Figure 1, B)–stained specimens. If available, LDs were confirmed using Sudan III staining (Figure 1, D). Furthermore, the presence or absence of BEFPD (Figure 1, E) and any findings indicative of diabetic nephropathy, including glomerular enlargement, mesangial sclerosis, basement membrane thickening, and arteriolar hyalinosis, were examined (Figure 1, F).¹⁴ 

Liver steatosis (Figure 2, A, C, and E), liver fibrosis (Figure 2, B, D, and F), cardiac fibrosis¹⁵  (Figure 3, A through D), coronary artery stenosis (Figure 3, E through H), and myocyte steatosis (Figure 3, I through L) were graded as shown in Table 1.

In the central nervous system, a diagnosis of acute Wernicke encephalopathy was based on swelling of the capillary endothelium, perivascular hemorrhages, relatively preserved neurons in the mamillary bodies, and/or hemorrhage and necrosis in the periaqueductal region of the brainstem. Chronic Wernicke encephalopathy was diagnosed on the basis of neuronal loss and gliosis in the mamillary bodies.¹⁶  In addition, pellagra encephalopathy was suspected when central chromatolysis was evident in neurons in the pontine nuclei (Figure 4, A).^17,18  Moreover, alcoholic cerebellar degeneration was suspected when cerebellar atrophy with a marked reduction in the number of Purkinje cells (<5 in 1 convulsion) and increased Bergmann glia was observed (Figure 4, B and C).¹⁸ 

During the autopsy, urine glucose and ketone bodies were measured by a dipstick using the nitroprusside method (Uropaper III, Eiken Chemical, Tokyo, Japan). Additionally, in all cases, drug intake was screened using the Triage DOA kits (Sysmex, Kobe, Japan), and ethanol concentrations in the blood and urine were measured using a gas chromatography–based autoanalyzer (GC-2014, Shimadzu, Kyoto, Japan). Furthermore, several whole blood, serum, and urine samples were collected at the autopsy. If DM was suspected to be the cause of death, the percentage of hemoglobin A_1c (HbA_1c) in the unfrozen blood sample and the concentration of serum autoantibodies, including anti–glutamic acid decarboxylase (GAD) antibody, anti-insulin antibody, anti–tyrosine phosphatase insulinoma antigen antibody, and anti–zinc transporter 8 antibodies, were measured. Other samples were immediately stored at −80°C for further examination. After the histopathologic confirmation of ketoacidosis, the concentrations of serum acetoacetate and BHB were measured. Serum BHB concentrations greater than 10.4 mg/dL (>1000 μmol/L; to convert to micromoles per liter, multiply by 96.05) were considered pathogenic.¹⁹ 

Immunohistochemistry for adipophilin (1:300–600 [kidney], 1:600–1000 [heart]; 2C5HB, Abcam plc, Cambridge, United Kingdom) was performed for the kidney and heart formalin-fixed, paraffin-embedded specimens using Leica Bond-IV automation and Leica Refine detection kits (Leica Biosystems, Bannockburn, Illinois) according to the manufacturer's instructions. Subsequently, all sections were counterstained with hematoxylin. The IHC staining patterns were compared with the results of Sudan III staining. Furthermore, 10 cases without ketoacidosis, a medical history of DM, or a possibility of alcohol abuse were used as controls (mean age, 44.4 ± 14.6 years; range, 22–70 years).

The diagnostic criteria of the pathologic background are shown in Table 2.²⁰  If there were several possible causes, we chose the most affected factor as the predominant cause considering the patient's circumstances, clinical information, and histopathologic findings.

Categorical data are presented as the number or percentage. Data were analyzed using IBM SPSS statistics (Version 26; SPSS Inc, Chicago, Illinois), and the significance level was set at .05. The Fisher exact test was used for categorical variables (causes of ketoacidosis and pathologic findings).

In addition, the Spearman rank correlation coefficient test was used for analyzing the relationship between serum BHB concentration and autopsy interval (between suspected time of death and autopsy) or measurement interval (between serum sampling and BHB concentration measurement).

A total of 16 cases exhibiting BV were identified (16 of 664; 2.4%). The clinical information and biochemical analysis results of these cases are summarized in Table 3. The case cohort included 11 men and 5 women (mean age, 53.8 ± 19.3 years; range, 26–92 years). A medical history of impaired glucose tolerance or DM was identified in 3 cases. Although a possibility of alcohol abuse was established in 8 cases, only 1 case had a premortem diagnosis of alcohol abuse. Combining clinical information and the histopathologic findings described below, the main cause of ketoacidosis was considered as alcohol abuse in 6 cases, DM in 5 cases, malnutrition in 3 cases, and hypothermia and infection in 1 case each. In cases with DM, 3 cases were diagnosed as autoimmune-mediated type 1 DM: case 12 showed elevation of serum anti-GAD antibody (6 U/mL; normal, <5 U/mL), anti–zinc transporter 8 antibody (21.8 U/mL; normal, <15 U/mL), and HbA_1c (12.7%; normal range, 4.6%–6.2%); case 15 showed elevated serum anti-insulin antibody (0.7 U/mL; normal, <0.4 U/mL) and HbA_1c (19.9%); and case 16 showed markedly elevated serum anti-GAD antibody (111 U/mL; normal, <5 U/mL) and HbA_1c (7.2%). All cases had a pathologic serum BHB concentration (mean, 51.6 ± 34.3 mg/dL [4960 ± 3296 μmol/L]; range, 10.7–114.5 mg/dL [1026–10 994 μmol/L]). The mean concentration of serum BHB was highest in the DM-predominant group (mean, 79.0 ± 35.3 mg/dL [7585 ± 3391 μmol/L]; range, 33.2–114.5 mg/dL [3189–10 994 μmol/L]), followed by the malnutrition-predominant group (mean, 52.7. ± 39.7 mg/dL [5063 ± 3816 μmol/L]; range, 10.7–106.0 mg/dL [1031–10 186 μmol/L]), and lowest in the alcohol-predominant group (mean, 31.0 ± 14.4 mg/dL [2979 ± 1384 μmol/L]; range, 10.7–55.1 mg/dL [1026–5292 μmol/L]). In all cases, the ketone body ratio (acetoacetate concentration/BHB concentration) was less than 0.05. The urine dipstick test was performed in 12 cases. However, ketone bodies were detected in only 2 cases (17%) in the DM-predominant group. Also, urine sugar was detected in 2 of the 6 patients (33%) with possible DM. Ethanol in the blood and/or urine was identified in 5 of 16 cases, all in patients who had a possibility of chronic alcohol abuse based on circumstantial information. No case showed positive results in the drug screening test.

In addition to the kidney, characteristic findings were mainly found in the liver, heart, and central nervous system. The autopsy findings, including heart weight and brain weight, and histopathologic findings and scores described above are summarized in Table 4. In the kidney, tissue degeneration was mild in 12 cases and moderate in 4 cases, and 13 of 16 cases (81%) showed BEFPD. Sudan III staining was performed in 14 cases, all of which showed positive results, regardless of the degree of tissue degeneration. Findings indicative of diabetic nephropathy were seen in 4 cases, but only 1 of these had a premortem diagnosis of DM. In the liver, high-grade steatosis was observed in 12 cases and high-grade fibrosis was identified in 6 cases. In particular, 5 of 6 alcohol-predominant cases showed both severe steatosis and moderate or higher-grade fibrosis. Furthermore, cirrhosis was observed only in this group. In the heart, severe fibrosis was observed in 3 patients, 2 of whom had DM and showed moderate or higher coronary artery stenosis grade, whereas the remaining patient had alcoholism and showed a moderate coronary artery stenosis grade. There were no cases with obvious pathologic findings in the conduction systems, including the sinoatrial node, the bundle of His, and the left and right branching bundles. In the central nervous system, central chromatolysis was identified in 2 cases (1 case each of alcohol and malnutrition). Moreover, findings consistent with alcoholic cerebellar degeneration were identified in 2 of 8 patients with a possibility of alcohol abuse. There were no findings suggestive of acute or chronic Wernicke encephalopathy or central pontine myelinolysis.

In the liver, although the steatosis grade was more severe in the possible alcohol abuse group than in the without/unclear alcohol abuse group, no statistically significant difference was observed between the 2 groups (Figure 5, A). Similarly, the fibrosis grade was more severe in the possible alcohol abuse group, and severe fibrosis (cirrhosis) was observed only in this group (Figure 5, B); however, no statistically significant difference was observed between the 2 groups regarding the total number of high-grade fibrosis and cirrhosis cases (P = .08 and .06, respectively). In contrast, a sum of steatosis and fibrosis grading of 5+ or more was found only in the possible alcohol abuse cases, a difference that was statistically significant (Figure 5, C; P = .03). In the heart, although both high-grade fibrosis (Figure 5, D) and high-grade coronary artery stenosis (Figure 5, E) were more common in the DM-positive group than in the DM-negative group, these differences did not reach statistical significance (P = .06 and .28, respectively). Regarding cardiac steatosis, high-grade steatosis was more frequently identified in the DM-negative group, and its frequency was significantly higher than that of the control cases (Figure 5, F; P < .001). Furthermore, severe steatosis was identified only in the possible alcohol abuse group, and the frequency of high-grade steatosis was significantly higher than in the control cases (Figure 5, G; P = .002).

The results of the IHC are shown in Table 4. In the kidney, microvacuolar staining patterns in the basal part of the tubules consistent with those of Sudan III staining were observed (Figure 6, A and B). This staining pattern was identified in 13 of 16 cases (81%). Immunoreactivity was obtained in all cases with mild tissue degeneration (12 of 12 cases), but in only 1 of 4 cases (25%) with moderate tissue degeneration, and the stainability was low and unclear (Figure 6, C). In control cases, only faint immunoreactivity was observed in the tubules, and the basal microvacuolar deposition pattern was not identified (Figure 6, D). In the heart, a microvacuolar staining immunoreactivity consistent with intracellular LDs observed in elastica-Masson and Sudan III staining was identified (Figure 6, E through G). However, immunoreactivity that appeared to be consistent with the cross-striation of the myocytes was identified in most cases, including the control cases (Figure 6, H), and this background immunoreactivity made it difficult to identify the former deposition pattern in most cases.

Both the autopsy and measurement intervals of all cases are summarized in Table 4, and the relationship between BHB concentration and postmortem or measurement interval is shown in Figure 7. The postmortem interval was estimated to be a maximum of 5 days (Figure 7, A; mean, 2.7 ± 1.3 days; range, 1–5 days), and there was no statistically significant relationship between BHB concentration and postmortem interval (P = .92). In contrast, the measurement interval ranged from 1 to 692 days (Figure 7, B; mean, 194.7 ± 192.8 days), and the difference was remarkable for each case. However, there was no statistically significant relationship between BHB concentration and measurement interval (P = .36).

As shown in previous studies,^1–4  the results of the present study showed that BV may be a reliable histopathologic marker of advanced ketoacidosis in postmortem investigations. Moreover, the serum BHB concentration was not affected by either the postmortem or measurement interval, consistent with previous studies.^8,9,21  Furthermore, BHB testing can help distinguish ketoacidosis from postmortem acetone/isopropanol production because it is a stable compound that shows no signs of postmortem formation.²¹  However, very high concentrations of BHB could be detected even after storage for more than 500 days, as seen in case 16. It has been reported that the amount of decrease in ketone bodies is smaller when stored at −80°C than at −20°C.⁹  Thus, it was considered important to store serum samples at −80°C for accurate diagnosis and further investigation of ketoacidosis.

In the present study, unlike in clinical practice, the nitroprusside method–based examination could not identify ketone bodies in most of the advanced ketoacidosis cases, excluding the DM-predominant cases.⁵  In contrast, BHB is the major ketone body produced in ketoacidosis, and the ratio between BHB and acetoacetate can rise from normal (1:1) to as high as 10:1 in diabetic and alcoholic ketoacidosis.⁷  This ratio was even higher in the present study, with the lowest rate being 25:1 and more than 40% of cases higher than 100:1. Another forensic autopsy–based study also showed this obvious ratio gap between the 2 ketone bodies.¹⁹  Laffel⁷  showed that the total amount of ketone bodies significantly increases in diabetic cases, whereas the increase is generally mild to moderate in malnutrition and alcoholic cases. Moreover, compared with BHB, acetoacetate is an unstable compound that is rapidly decarboxylated.⁹  In ketoacidosis cases, excluding DM-associated cases, a milder increase in the total amount of ketone bodies and/or rapid postmortem decarboxylation of acetoacetate might be associated with the high prevalence of negative reactions by urine dipstick examination. This study showed that the prevalence of ketoacidosis may be higher than expected, and detecting ketoacidosis at autopsy may be difficult in not a few cases. Appropriate correction of the death certificate may be required when unexpected ketoacidosis is retrospectively evaluated as a major contributing factor in the cause of death.

This study demonstrated that the presence of BV was important for the diagnosis of ketoacidosis in a postmortem investigation, even if suggestive clinical signs were not evident. In addition, IHC for adipophilin was found to be a reliable histologic marker of BV, as was fat staining using frozen sections. Alternatively, we should note that IHC for adipophilin was not helpful in the heart, and the stainability was not uniform in the other organs, especially in cases showing moderate tissue degeneration. Both careful organ selection and estimation of the degree of postmortem degeneration may be required when using this antibody to diagnose ketoacidosis with BV. Moreover, LD accumulation in the renal tubules could be associated with other pathologic conditions, including hyperlipidemia²  and Reye syndrome.²²  Furthermore, in DM patients, Armanni-Ebstein lesions, which may be confused with BV, can also be observed.^23–25  Thus, confirmation of BHB concentration is essential to distinguish ketoacidosis from these conditions.

In all alcohol-predominant cases, severe steatosis accompanied with moderate or higher fibrosis was observed in the liver. In addition, cerebellar degeneration suggestive of alcoholic cerebellar degeneration was observed in a quarter of the cases. These findings were easily recognized even under macroscopic observation and considered helpful for suspecting premortem alcohol abuse. Furthermore, the presence of central chromatolysis, suggestive of pellagra encephalopathy, in this group indicates the importance of nicotinamide in the treatment of alcoholic ketoacidosis.

High-grade steatosis in the heart was more frequently observed in cases with possible alcohol abuse. The formation of LDs and increased fat accumulation in the cardiomyocytes have been associated with alcoholic cardiomyopathy in human cardiac tissues²⁶  and mouse models.^27,28  In addition, patients with alcohol abuse often suffer from type 2 DM and malnutrition,^29,30  both of which also have been reported to cause intracytoplasmic LD accumulation.^28,31–33  In the present study, 2 cases showing severe steatosis had malnutrition (case 3) or type 2 DM (case 7) in addition to the possibility of alcohol abuse; therefore, the combination of these factors may be associated with significant cardiac steatosis. We should also note that LD accumulation has been associated with lethal arrhythmias.^34,35 

In the DM-predominant group, the low prevalence of cases with a clinical diagnosis and pathologic findings suggestive of diabetic nephropathy may suggest that disease duration may not be associated with the onset of ketoacidosis. Furthermore, in the present study, ketoacidosis associated with autoimmune-mediated type 1 DM was a significant cause of unexpected death in adults less than 40 years old (3 of 5 cases). A previous clinical study showed that type 1 DM patients less than 40 years of age have a 4- to 10-fold rate of sudden and unexpected death compared with the general population; thus, this condition should be always considered as one of the possible causes of death in this population.^36–38  Both postmortem pathologic and serologic investigations targeting DM should be emphasized, even in younger victims.

The present study was limited by the clinical information of some patients—mainly the lack of severe clinical symptoms or low medical institution consultation rates. In addition, the blood HbA_1c level and the histopathology of the pancreas could not be assessed in some cases because of the progression of postmortem change. Furthermore, although acetone is a useful parameter for diagnosing ketoacidosis,²¹  we could not detect acetone in most samples, possibly owing to the limit of quantification adopted in our laboratory (approximately 15 mg/dL). Moreover, we do not routinely sample vitreous fluid during autopsy and could not evaluate vitreous glucose concentrations.³⁹  Detailed histopathologic findings in hypothermia or infection-predominant ketoacidosis remain undetermined because of a paucity of such cases. Finally, we did not investigate the cases of ketoacidosis without BV, if such cases were present in our autopsy cases.

The present study suggests that BV may be a histopathologic hallmark of advanced ketoacidosis, and BHB is a suitable compound not only for its definitive diagnosis but also for the retrospective analysis of advanced ketoacidosis. In contrast, the nitroprusside method–based dipstick examination could not identify ketone bodies in most cases in the autopsy setting. Careful histopathologic examination is important for determining the background causes of ketoacidosis with BV, especially DM and chronic alcohol intoxication. In addition to traditional fat staining, IHC for adipophilin is a useful alternative for demonstrating fat accumulation in BV.

The authors are grateful to Noboru Onozuka, MT, Syuko Matsumori, MT, Miyuki Maekawa, MT, and Osamu Yamamoto, MT, for their technical assistance.