Emphysematous pyelonephritis: experience at a tertiary care hospital in Bangladesh

The diagnosis of EPN was confirmed by abdominal CT scan. One patient had EPN in an ectopic right kidney and one had EPN with a psoas abscess. The right kidney was involved in 9 patients (45%) and the left in 11 patients (55%). Two had evidence of pyelonephritis in contralateral kidneys along with EPN.

According to the Huang and Tseng classification,3 three patients (15%) had class 3B EPN (Figure 1), one (5%) had class 3A, and 16 (80%) had class 2 (Figure 2).

Figure 2 Non-contrast CT scan of the abdomen (axial film) showing class 2 (gas within renal parenchyma) emphysematous pyelonephritis of the right kidney (wide arrow) with gross parenchymal destruction

figure_2.jpg

Urine culture revealed the growth of microorganisms in 16 patients (80%). Escherichia coli was the most common organism identified on urine culture (12; 60%), of whom five (41.7%) had extended-spectrum beta-lactamase (ESBL)-positive E. coli; three of these cases (25%) were complicated by bacteraemia. Blood appeared to be sterile in all other cases. Less common microorganisms were Klebsiella pneumoniae (1 patient; 5%), Candida albicans (1; 5%), non-albicans Candida (1; 5%) and mixed bacterial species (1; 5%). One patient had sterile urine and blood, but culture of pus collected during nephrectomy revealed growth of ESBL-positive E. coli.

All patients were treated with intravenous antibiotics and other supportive measures including fluid resuscitation. Two patients underwent a few sessions of haemodialysis. Four (20%) required surgical interventions (nephrectomy in 3; 15%) and open drainage (1; 5%). The requirement for nephrectomy was associated with a higher radiological class of EPN (p = 0.034) and AKI (p = 0.032) (Table 2). No patient required percutaneous drainage or double-J (DJ) stenting. Biopsies of nephrectomised tissues revealed evidence of acute-on-chronic nephritis with micro-abscess formation. The patients were hospitalised for 7–31 days. AKI resolved in 8 patients before discharge and there were no deaths. The management and outcomes are shown in Table 3.

Discussion

In 1898, Kelly and MacCallum were the first to report a case of gas-forming renal infection (pneumaturia).⁹ Since then, various terms like ‘pneumonephritis’ and ‘renal emphysema’ have been used to describe this entity. The term ‘emphysematous pyelonephritis’ was adopted by Schultz and Klorfein in 1962.^2,10

EPN predominantly occurs in females, and most commonly involves the left kidney, followed by the right kidney; bilateral involvement is the least common.^3,7,11 There are exceptional cases of bilateral renal involvement (class 4) and in which concomitant emphysematous cystitis has been reported.¹²No patients in our series had bilateral (class 4) disease, in contrast to the study of Misgar et al. in which a third of their EPN patients had bilateral involvement.¹¹ However, we noted bilateral renal infection in two cases: EPN on one side and pyelonephritis on the other side, which aligns with Karthikeyan et al. who found it in one-tenth of their cases.¹³ It is not known why EPN predominantly involves the left kidney, or why it occurs on one side and pyelonephritis in the other in the same patient.

The mechanism of gas formation in EPN is widely attributed to glucose fermentation by enteric bacteria.3,14 The pathogenesis of EPN may also include infection by gas-forming organisms, tissue ischaemia, anaerobic environments and defective immune mechanisms.3 Vesicoureteric reflux also contributes.4 Obstruction of renoureteral units by stones, papillary necrosis or tumours can predispose to EPN and stones may act as a nidus for infection.^1–3,5

Patients with EPN present with the typical features of an upper urinary tract infection (e.g. fever, renal-angle pain or tenderness, and vomiting).^1–3,6,7 Occasionally they have altered consciousness and shock.^3,14 Virtually no feature is diagnostic of EPN. The clinical presentation of fever, loin pain and vomiting in our series is comparable with other published series.^11,13–16 Asymptomatic cases are occasionally identified during investigation for some other reasons.¹⁷

Neutrophilic leucocytosis, high ESR, high CRP and poor glycaemic status are consistent features as most EPN cases have DM.^{1–3,13–16} Occasionally an episode of EPN unmasks previously undiagnosed DM,^3,18 which happened with one of our patients. E. coli is the most commonly isolated organism in most series.^{1–3,6,7,13–16} In our series, two patients had fungal growth (one case each of C. albicans and non-albicans Candida) on urine culture, but their blood remained sterile. Both were put on antibiotics after sending their urine and blood for cultures (as per hospital protocol); as a result of their clinical responses and urine culture reports, we did not add antifungals. We assumed that the fungal growth was facilitated by indwelling catheters. EPN caused by Candida has been treated with antifungals and nephrectomy.¹⁹ A fifth of our patients had no urine isolates, which is not unusual, especially in patients who receive antibiotics before seeking medical care,¹ or who are transferred from other centres where they had been treated. However, ESBL-positive E. coli was cultured in pus collected during nephrectomy from one patient.

Bacteraemia, AKI and electrolyte imbalances were important complications in our series. In other studies, up to 75% of EPN cases were complicated by AKI.^1,2 Our patients stayed in hospital for 7–31 days and most of them received antibiotics for 4 weeks, which was consistent with another report.¹ None of our patients required transfer to the intensive care unit (ICU), which may relate to the predominantly less severe cases (class 2; 80%) in our series. Two patients with class 3B EPN (Figure 1) and one with class 2 EPN with gross renal parenchymal destruction (Figure 2) required nephrectomy. Huang and Tseng concluded that class 1, 2 and less aggressive presentations of class 3 and 4 EPN cases (with 1 out of 4 risk factors including shock, altered sensorium, AKI and thrombocytopaenia) may be managed by antibiotics and percutaneous drainage, while fulminant presentations of class 3 and 4 cases (2 or more risk factors) merit nephrectomy.3 Successful medical management of severe class 4 EPN is reported and many investigators nowadays use nephron-sparing strategies with favourable outcomes.^20,21

Mortality associated with EPN has reduced largely in the last two or three decades. Poor prognostic features include altered conscious level and shock during admission, haematuria, proteinuria, hypoalbuminaemia, thrombocytopaenia, uraemia and the requirement for haemodialysis, polymicrobial infection, bacteraemia and extension of infection to perinephric areas.^2,3 Our series had zero mortality, although some risk factors were present, such as low platelet count, hypoalbuminaemia, altered sensorium, uraemia and the requirement for haemodialysis. Irfaan et al. reported a series of 20 EPN patients, of whom nearly half had altered sensorium and a quarter had shock, without any mortality.14 Many factors contribute to the improved outcomes in EPN, including wide availability of CT scan facilities, early detection and less advanced radiological classes of EPN, aggressive resuscitative measures, rapid glycaemic control using intravenous insulin, early administration of effective, broad-spectrum intravenous antibiotics, interventional radiology, minimally invasive intervention techniques1 and a multidisciplinary team approach.²

Our study had some limitations: a small number of patients, retrospective evaluation, and location in a single centre. However, most other EPN series also have relatively small patient numbers and mostly evaluate hospital records retrospectively.^{1,2,5,7,11,14} A protocol-based, multicentre, prospective study with a larger sample may inform clinical practice and help establish management guidelines for EPN patients.

In conclusion, EPN occurred predominantly in females in this study, and all patients were diabetic. Fever, loin pain, vomiting and dysuria were common. Two-thirds of patients had AKI and a fifth required surgical intervention, including nephrectomy. There were no deaths. Although conservative management of EPN shows promising outcomes, we emphasise that interventions should not be delayed (as and when indicated) and may be life-saving. A high index of suspicion is warranted in appropriate common clinical scenarios, especially in patients with DM who respond slowly to antibiotic treatment for pyelonephritis, so that a CT scan can be performed early and an appropriate management strategy instituted to improve outcome.

Acknowledgement

The authors would like to express their thanks to Samira Humaira Habib, Senior Research Officer, Health Economics Unit, Bangladesh Diabetic Somiti (BADAS), Dhaka, Bangladesh for analysing the data.