Association Between Causative Pathogen and Occurrence of Infection-Related Glomerulonephritis in Infective Endocarditis

Abstract

Infection-related glomerulonephritis (IRGN) has long been recognized as a complication of infective endocarditis (IE) [1, 2]. The prevalence of IRGN was very high in the preantibiotic era, reportedly in excess of 75% in some series [2, 3]. Its incidence was found to be much reduced in the setting of antibiotic treatment [4], and glomerulonephritis was not even mentioned in a large multicenter study on presentation of IE in the 21st century [5]. Nevertheless, the condition is still encountered in centers where large numbers of patients with IE are treated and remains an important complication of the disease.

In a series of 49 cases of IRGN diagnosed by means of kidney biopsy between 2001 and 2011, the majority of cases occurred in patients with IE caused by Staphylococcus aureus (53%) or Streptococcus spp. (23%) [6]. In this series, 4 (8%) of the 49 cases had IE caused by Bartonella henselae [6]. Bartonella is otherwise an uncommon cause of IE. In a large multicenter prospective cohort study of IE cases admitted to 58 hospitals in 25 countries between June 2000 and September 2005, <1% of IE cases were caused by Bartonella spp [5]. Bartonella IE is similarly uncommon in our institution, accounting for <1% of IE cases. However, we have noticed that an unusually large proportion of these patients have IRGN. The purpose of the current study was to examine the strength of the association between infection with various pathogens and the occurrence of IRGN in IE.

METHODS

The study was designed as a nested case-control study among patients with pathologically proven IE. Episodes of IE for patients admitted to Cleveland Clinic from 1 July 2007 through 31 December 2021 were identified from the Cleveland Clinic Infective Endocarditis Registry and screened for inclusion in the study. This registry is populated by screening all patients operated for a diagnosis of IE, all patients in the Cleveland Clinic Outpatient Parenteral Antimicrobial Therapy Registry with a diagnosis of IE, and all patients who died in the hospital with a diagnosis of IE. The study was approved by the Cleveland Clinic Institutional Review Board as exempt human subjects research (IRB no. 19-098).

Only adults with pathologically proven IE were included. Those with polymicrobial IE, those without an identified pathogen, and those with end-stage renal disease were excluded. Also excluded were non-Bartonella IE episodes for patients with Bartonella IE and subsequent IE episodes for patients already included.

Definite IRGN was defined as histopathological evidence of acute proliferative glomerulonephritis on a kidney biopsy. Probable IRGN was defined as acute kidney injury (≥1.20 mg/dL, which was 25% higher than the upper limit of the reference range for serum creatinine in our laboratory) associated with hematuria, proteinuria, and depleted complement C3 and/or C4 levels. Possible IRGN was defined as either the presence of hematuria, proteinuria, and acute kidney injury or the presence of hematuria, proteinuria, and depleted complement levels. Injection drug use was defined as having injected illicit drugs within 3 months before admission. Infection of a prosthetic valve or prosthetic ring was categorized as prosthetic valve endocarditis (PVE).

Patients with definite or probable IRGN were considered case patients. Four controls were selected for each case patient from patients without definite, probable, or possible IRGN, by propensity score matching on age, sex, admission year, and IE category (native or PVE), using the nearest-neighbor method and the R software package MatchIt [7].

The association between each pathogen and the occurrence of IRGN was examined in separate logistic regression models. The analysis was performed by one of the authors (N. K. S.) using R software, version 4.2.2 [8].

RESULTS

Forty-eight case patients (IE with definite or probable IRGN) were identified. Seventeen of the 48 case patients (35%) had definite IRGN; the remaining 31 had probable IRGN. The matching process found 192 controls (IE without IRGN). There were no significant differences in the distributions of age, sex, proportion with PVE, and proportion with injection drug use between case patients and controls (Supplementary Table 1).

There was a significant difference in the distribution of pathogens between case patients and controls. Bartonella was the causative pathogen in 8 (17%) of the case patients but only 1 (.5%) of the controls. Although representing only 3.75% of the study participants, 8 of the 9 patients with Bartonella infection had IRGN.

Among the 48 case patients with IRGN and 192 controls without, the odds of a study participant with Bartonella IE having IRGN was significantly higher than those of a participant with non-Bartonella IE having IRGN (8:1 vs 40:191, respectively; odds ratio, 38.2 [95% confidence interval, 6.7–718.8]; P < .001). There was no significant association of IRGN with IE caused by S. aureus, viridans group streptococci, Enterococcus, coagulase-negative staphylococci, other bacteria (grouped), or fungi (Table 1).

DISCUSSION

The current study demonstrates a very strong association between Bartonella IE and the occurrence of acute glomerulonephritis. The magnitude of the effect makes it unlikely that the observed association was a spurious one. Including only patients with pathologically proven IE ensured that all cases of IE were IRGN and not acute glomerulonephritis unrelated to IE; this could have happened if IE were defined using the widely accepted Duke criteria [9], which include acute glomerulonephritis itself as a minor criterion for the diagnosis of IE. Excluding patients with possible IRGN should have limited the possibility of mistakenly including those with a missed diagnosis of IRGN among the controls.

A limitation of the study is that not all cases of IRGN were pathologically proven, but this limitation cannot be overcome, as a reasonable risk-benefit calculation during routine care will not allow kidney biopsy in most patients with suspected IRGN. However, given the strict definition for probable IRGN, it is unlikely that the case patients included anyone without IRGN. Another limitation is that the duration of illness was not controlled for. It is not possible to get an accurate duration of illness for all patients in a retrospective study conducted in a referral center where patients present with the diagnosis already made and where the emphasis is on treating the condition for which they were referred. Because of this limitation, it is not clear whether IRGN is related to a prolonged duration of untreated IE or secondary to a pathogen-specific pathogenetic process.

Most patients have residual renal impairment after an episode of IRGN. Among 30 survivors in an IRGN series, 4 (13%) progressed to end-stage renal disease, 14 (47%) had persistent renal dysfunction, and only 12 (40%) had complete renal recovery [6]. Although IRGN was initially thought to be an embolic complication, its pathogenesis is now considered to involve immune complex–mediated injury [10–12]. This suggests that the most effective treatment for IRGN would be to eliminate the inciting cause, which is IE. Early diagnosis and appropriate treatment of IE would be expected to improve outcomes of IRGN.

Unfortunately, because Bartonella does not grow in routine blood cultures, the diagnosis of IE in patients with Bartonella IE is usually delayed. Over the years we have seen several patients who were treated with immunosuppressive therapy for acute glomerulonephritis without an appreciation that their condition was a manifestation of underlying IE. Recognition of the very strong association between Bartonella IE and acute glomerulonephritis should lead to consideration of Bartonella IE as an underlying diagnosis in patients with a diagnosis of acute glomerulonephritis. In addition, acute glomerulonephritis diagnosed in a person with clinical or echocardiographic features suggestive of IE, along with negative blood cultures, should prompt specific testing for the presence of Bartonella infection by means of blood PCR or serology.

In conclusion, Bartonella IE is very strongly associated with IRGN. Recognition of this association may expedite diagnosis and appropriate treatment of both the IE and the IRGN in afflicted patients.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Author Contributions. Conceptualization, data curation, software, formal analysis, visualization, writing—original draft preparation, and supervision: N. K. S. Methodology: N. K. S., G. N. N., and L. C. H. Validation and investigation: N. K. S. and E. C. K. Project administration: N. K. S. and S. M. G. Resources: S. M. G. Writing—reviewing and editing: All authors.

References

Author notes