Severe SARS-CoV-2 infection is associated with a high incidence of acute kidney injury (AKI), ranging in different studies between 4.5% [1] and 28% [2], with a higher prevalence in patients hospitalized in the intensive care unit (ICU) [1, 2]. In addition, among patients with coronavirus disease 2019 (COVID-19) hospitalized in the ICU, AKI was an independent risk factor for mortality [3].

Along with AKI, other laboratory patterns of kidney damage, such as the presence of isolated urinary abnormalities like proteinuria and haematuria, have been described in patients affected by COVID-19, albeit less frequently [4].

Although several mechanisms have been proposed to explain the pathogenesis of COVID-19–associated AKI [4,5,6], a better understanding of kidney pathology findings in COVID-19 still represents an unmet need, and reports of autopsy studies are very limited in the literature.

In this single-centre retrospective study, we examined the kidney samples of 43 patients with a severe form of COVID-19 who died in the ICU of the Ospedale Luigi Sacco, Milan, Italy, during the first COVID-19 wave of 2020. The total number of ICU admissions during the enrolment period was 101, and 43 of the 54 patients who died underwent autopsy.

Biopsies were performed on both kidneys shortly after death, following which, tissues were fixed in 10% buffered formalin at room temperature for up to 48 h, and histopathological examination was performed using tissue sections stained with periodic acid-Schiff, acid fuchsin orange trichrome, silver stain, and hematoxylin–eosin. The extent of glomerular and tubular damage was classified according to the score described in Table 1.

Table 1 Grading of tubular and glomerular damage
Full size table

To assess tubular damage, we also considered the presence of endoluminal casts and the erosion of the tubular epithelium. Finally, to grade the glomerular damage we evaluated the general appearance of the glomerular flocculus.

In addition to histopathological examination, molecular biology analysis for the detection of SARS-CoV-2 was performed on the renal tissues of 20 patients (46%).

RNA was extracted from 5 μm thick paraffin-embedded sections using Quick- RNA FFPE Miniprep (Zymo Research). Two SARS-CoV-2 target genes were then amplified and tested in all patients: (i) RNA-dependent RNA polymerase gene (RdRP gene, Co-Diagnostic, Salk Lake City, Utah, USA), (ii) envelope (E gene, the WHO/Charité, Berlin, Germany).

The mean age of the examined population was 63.2 years (range min–max 34–80, SD±9.1), and 93.0% of the patients were male. Mean hospitalization time was 12.5 days (SD ±9.7).

AKI was frequent in our population, affecting 24 patients (55.8%). Eleven of the patients with AKI (47.8%) needed continuous renal replacement therapy. As in other forms of acute respiratory distress syndrome, patients with COVID-19 acute respiratory distress syndrome underwent numerous cycles of position change from supine to prone, according to their arterial blood gases evaluated by pO2/FIO2 ratio. Fifteen (65%) patients underwent position change from supine to prone whereas 15 (65%) underwent prone-supination at least once in the 72 h before death. All patients required invasive ventilation, with administration of different levels of Positive End Expiratory Pressure (PEEP). The mean PEEP value administered in the first 7 days of admission to the ICU was 13.77 cmH2O (range min–max 4.85–18, SD±3.11).

We identified several patterns of histopathological changes in the renal parenchyma. The tubules showed severe alteration of the epithelium, with denudation of the basement membrane, loss of the brush border, and eventually increased cellular mitosis. Moreover, we identified the presence of casts within the lumen of the tubules, with a rigid, stiff, and dichromic appearance.

On the other hand, glomerular damage consisted mainly of the collapse of the glomerular tuft on the vascular pole (Fig. 1). Of note, none of the examined cases showed sclerosis of the mesangium, or necrosis and inflammatory changes, and neither the interstitium nor the glomerulus showed significant inflammatory infiltrate. Indeed, no lympho-monocytic inflammatory infiltrate or vascular and thrombotic lesions were observed in any of the cases.

Fig. 1
figure 1

Representative findings of microscopic evaluation of kidney biopsies collected shortly after death in COVID-19 patients admitted to the intensive care unit. Panel A. Low magnification of renal parenchyma with diffuse glomerular damage. (Periodic acid-Schiff (PAS) staining, OM 10x). Panels B, C. Grading of the glomerular damage: glomerulus with minimal collapse of the flocculus, in which the capillary loops are still visible (panel B), glomerulus with severe collapse of the flocculus on the vascular pole and Bowman’s space characterized by the presence of corpuscular material and cellular debris (panel C). (PAS staining, OM 20x). Panels D, E. Tubular damage. In all panels, the lumen of the tubules are occupied by irregular casts) and amorphic debris; the tubular cells are partly necrotic and denudation of the basement membrane is observed. (Masson Trichome and PAS staining, OM 40×)

Full size image

In our population, 55.8% of patients (24/43) had mild tubular damage, 39.5% (17/43) had moderate tubular damage, and 4.65% (2/43) had severe tubular damage (Fig. 2A).

Fig. 2
figure 2

A–B Grading of histological damage in the population

Full size image

Glomerular damage was more severe, with 39.5% of patients (17/43) having mild glomerular damage, 37.2% (16/43) with moderate glomerular damage, and 23.2% (10/43) with severe glomerular damage (Fig. 2B). Comparison between the different degrees of glomerular and tubular damage revealed no differences related to the age of the patients (Table 2).

Table 2 Age (mean ± SD) of patients according to different degrees of glomerular and tubular damage. No significant differences were detected (one-way ANOVA)
Full size table

Molecular evaluation for the identification of the SARS- CoV-2 genome on the kidney tissues by polymerase chain reaction was positive in only 4/20 cases (20%), equally distributed between the AKI group and the non-AKI group.

Acute tubular damage is the main risk factor leading to AKI. In our study, moderate and severe tubular damage was present in 19 patients and correlated with AKI [OR = 54 (95% CI 5.9–495.0) p = 0.0004]. However, our study also revealed the presence of severe glomerular damage. Indeed, collapsing glomerulopathy was detected in a high percentage of patients (26/43–60.4%) and significantly correlated with the presence of AKI [OR 6.5 (95% CI 1.7–25.3); p = 0.007].

Collapsing glomerulopathy is not a typical pattern described for AKI. In addition, the morphological appearance of the glomerular damage was consistent with new-onset ischaemic damage, suggesting that the glomerular damage might be the consequence of the ischaemic insult itself.

Interestingly, no signs of direct viral damage, such as interstitial inflammatory infiltrate were detected.

Based on the observation of non-specific histological findings, we hypothesise that kidney abnormalities might be the consequence of multifactorial damage, mainly produced by the ischaemic injury. In addition, we previously documented a significant correlation between PEEP levels and AKI in COVID-19 patients, which could contribute to develo** this peculiar type of kidney damage [7].

However, despite the observed severe tubular and glomerular injury, the absence of sclerosis suggests potentially reversible ischaemic damage. Kidney biopsies from patients who experienced COVID-19-related AKI and eventually recovered would help understand the long-term kidney outcome.