The effects of add-on corticosteroids on renal outcomes in patients with biopsy proven HIV associated nephropathy: a single centre study from South Africa

Despite the era of antiretroviral therapy [ART], human immunodeficiency virus [HIV]–related kidney disease still carries a significant risk of end-stage renal disease (ESRD) [1]. From available data in Africa the prevalence of chronic kidney disease (CKD) in HIV-infected patients is reported to be 38% in Nigeria [2], 33.5% in Zambia [3], 20% in Uganda [4], 11.5% in Kenya [5], and 5.5–6% in South Africa [6, 7]. The regional variations in CKD prevalence is likely due to differences in reporting methods, sampling, definition of CKD, access to health care and genetic factors [8].

The recent [Kidney Disease Improving Global Outcomes] KDIGO Controversies Conference publication, endorsed a pathological classification of HIV-related kidney diseases. The classification encompassed glomerular-, tubulointerstitial- and vascular-dominant lesions as well as other lesions seen in the setting of HIV [9]. Classic HIV associated nephropathy [HIVAN] is defined as a collapsing glomerulopathy with interstitial inflammation, microcysts and tubular injury [9]. The interstitial component includes a diffuse inflammatory cell infiltrate which may include plasma cells [8, 10, 11, 12, 13]. Numerous studies describe the release of several inflammatory mediators e.g. interleukin 6, nuclear factor kappa B [14] and transforming growth factor β [15]. These mediators promote interstitial inflammation and subsequent fibrosis [14, 16, 17].

Mortality is reduced when initiating antiretroviral therapy [ART] in patients with HIVAN [11, 18]. However despite ART improving mortality and renal impairment there are still a subgroup of patients where CKD progresses regardless of ART [11]. This may be explained by the kidneys acting as a viral reservoir for HIV nucleic acid, even on ART [19, 20, 21].

A number of old studies have evaluated the efficacy and safety of corticosteroids in the treatment of classic HIVAN [22, 23, 24]. However, these studies were limited by their retrospective, non-randomized design together with different ART and steroid regimens. There has also been safety concerns due to increased risk of infections [18]. To the best of our knowledge only one retrospective study of 19 patients has assessed the use of corticosteroids for the treatment of HIVAN in the triple therapy era of ART and demonstrated a significant improvement in estimated glomerular filtration rate [eGFR] in the corticosteroid arm [24]. Further trials to assess corticosteroid benefit have therefore been recommended [18, 21].

The aim of our study was to assess the balance between improvement in eGFR, proteinuria and adverse events with adjuvant corticosteroids to ART in patients with HIVAN. Our hypothesis was that there would be a greater improvement in eGFR in the corticosteroid cohort due to the dampening effect of inflammation in the interstitial compartment in those patients with HIVAN. The importance is underscored by the high prevalence of CKD in HIV positive patients in Africa and the lack of prospective treatment studies.

This was a prospective, open labelled trial to investigate the response to corticosteroids as adjuvant treatment to ART in patients with glomerular and interstitial features of HIVAN. The study was approved by the Human Research Ethics Committee of the University of Cape Town [HREC:241/2007]. The trial was registered with the ISRCTN registry [ISRCTN study ID (56112439].

A total of 1131 renal biopsies were performed during the trial period. Of these, 354 [31.3%] were from HIV-infected patients (Fig. 1: Study flowsheet for the trial). All patients had a clinical indication for biopsy. There were 38 patients who met criteria for inclusion into the trial; of these, 21 patients (55.3%) were treated with [ART+C] while 17 patients (44.7%) were treated with [ART Alone]. Table 1 describes the baseline characteristics of the cohort.

Open image in new window

Table 1

Baseline demographic, clinical and biochemical parameters

	ALL N = 38	[ART + C] N = 21	[ART Alone] N = 17	P Value*
Demographics
Gender (female), n[%]	25 [65.8%]	14 [66.7%]	11 [64.7%]	0.899a
Age (years) median IQR	33 [2;39]	36 [27;42]	28 [25;34]	0.050b
Clinical parameters
Systolic BP (mmHg) median [IQR]	118 [110;125]	118 [110;125]	118 [110;121]	0.668b
Diastolic BP (mmHg median [IQR]	77 [71;83]	76 [65;80]	77 [71;84]	0.561b
Diabetes; n [%}	1 [2.6%]	1 [4.8%]	0	1.000a
Hypertension, n [%}	7 [18.4%]	5 [23.8%]	2 [11.8%]	0.306a
Tuberculosis at baseline, n [%}	10 [26.3]	7 [33.3%]	3 [17.6%]	0.242a
Commenced on ACE/ARB; n [%}	27 [71.1%]	14 [66.7%]	12 [70.6%]	0.519a
ART regimen at baseline, n [%}				0.332a
- D4T/3TC/EFV or NEV	20 [52.6%]	11 [52.4%]	9 [52.9%]
- ABC/3TC/EFV or NEV	9 [23.7%]	7 [33.3%]	2 [11.8%]
- ZVD/3TC/EFV or NEV	6 [15.8%]	2 [9.5%]	4 [23.5%]
- Other/unknown	3 [7.9%]	1 [4.8%]	2 [11.8%]
Biochemical parameters
eGFR (ml/min/1.73m2) median[IQR]	43.5 [26;70]	35 [18;46]	47 [39;97]	0.015b
eGFR (ml/min/1.73m2) mean [SD]	50 [32.8]	37.8 [26.0]	65.1 [34.8]
CD4 (cells/mm3) median[IQR]	158 [63;309]	158 [36;272]	173 [104;352]	0.437b
HIV VL (Log 10) (cps/ml) median[IQR]	4.77 [3.87;5.29]	4.77 [4.5;5.29]	4.69 [2.75;5.3]	0.5623b
Albumin (g/L) median[IQR]	30 [25;33]	29 [24;32]	30.5 [27;36]	0.262b
uPCR (g/mmol) median[IQR]	0.36 [0.14;0.54]	0.43 [0.17;0.52]	0.23 [0.13;0.60]	0.617b
Haemoglobin (g/dL) median[IQR]	8.1 [7.3;10.2]	8.0 [7.4;9.9]	8.1 [7.3;10.2]	0.872b
Cholesterol(mmol/l) median[IQR]	4.5 [3.6;5.6]	4.4 [3.3;5.1]	4.8 [3.9;6.0]	0.347b

ART + C Antiretrovirals + corticosteroids, ART Alone Antiretrovirals alone, IQR interquartile range, BP Blood pressure, ACEi angiotensin converting enzyme inhibitor, ARB angiotensin receptor blocker, D4T Stavudine, 3TC Lamivudine, NEV Nevirapine, ABC Abacavir, EFV Efavirenz, ZVD Zidovudine, uPCR urine protein-to-creatinine ratio, HIV VL HIV viral load, eGFR estimated glomerular filtration rate, SD standard deviation. a = Fishers Exact test, b = Wilcoxon Rank-Sum test, * p-value comparison of [ART+C] vs [ART Alone]

Overall, there were more females (65.8%), median age was 33 (IQR: 25–39) years, median SBP and DBP was 118 (110–125) mmHg and 77 (71–83) mmHg, respectively. All patients included in the trial were of black African ethnic origin. Only one patient (2.6%) had diabetes while 7 [18.4%] were known with hypertension. There were ten patients (26.3%) who were diagnosed with TB at time of presentation. Patients were initiated on an ACE-I or ARB if their BP could tolerate it and if hyperkalaemia was not a complication. Overall 71.1% of patients were treated with an ACE-I or ARB ([ART +C] 14/21 and [ART Alone] 12/17). All patients received triple regimen ART therapy. The most common initial ART regimen [52.6%] consisted of stavudine (D4T), lamivudine (3TC) and efavirenz (EFV) /nevirapine (NEV). The other ART combinations are detailed in Table 1; no patient received tenofovir.

Changes in clinical and biochemical parameters from baseline until last follow–up

All patients were analyzed as per intention to treat for the primary outcome. Additional file 1 Table S1 includes all changes in median CD4 and viral load at 3-,6-, 12- and 24-months. Over the study period the eGFR improved in both the groups (Fig. 2: eGFR by duration on ART with and without the addition of corticosteroids.) Figure 3 describes the median eGFR and IQR at 6, 12- and 24-month periods. The group receiving [ART+C] had a statistically significant improvement in median eGFR from baseline to last follow-up compared with [ART Alone] i.e. [Δ = 25mls/min (IQR: 15–51) vs 9 mls/min (IQR: 0–24), p = 0.008] (Table 3).

Open image in new window

Open image in new window

Table 3

The change [Δ] in clinical and biochemical parameters at 24 months

Median change [Δ] in parameters at 24 months [IQR]	[ART + C]	[ART Alone]	P Valuea
Clinical parameters
Systolic BP mmHg [IQR]	10 [−25; 11]	6 [2; 18]	0.466
Diastolic BP mmHg [IQR]	−6 [−20;4]	1 [−5; 7]	0.177
Biochemical parameters
Change in eGFR (ml/min/1.73m2) [IQR]	25 [15;51]	9 [0;24]	0.008
Change in uPCR (gm/mmol) [IQR]	−0.13 [−0.25; −0.08]	−0.12 [−0.55; −0.06]	0.691
Change in Haemoglobin (gm/dL)[IQR]	3.4 [0.3; 4.7]	3.6 [2.1; 4.8]	0.480
Change in Albumin (gm/L)[IQR]	5 [2; 12]	6 [4;15]	0.445
Change in Cholesterol (mmol/l)[IQR]	0.2 [−0.10; 1.7]	0.3 [−1.5; 0.9]	0.577

ART + C Antiretrovirals + corticosteroids, ART Alone Antiretrovirals alone, BP Blood pressure, eGFR estimated glomerular filtration rate, uPCR protein creatinine ratio, a = Wilcoxon Rank-Sum test

Additional file 2 Table S2 describes all eGFR values at baseline and last follow-up. Proteinuria improved in both groups over the trial period however the change in proteinuria was not statistically significant between the groups. [(ART+C) Δ = − 0.13 g/mmol (− 0.25;-0.08) versus (ART Alone) Δ = − 0.12 g/mmol (− 0.55;0.06) p = 0.691] (Table 3). There was no clinically significant change in BP, Hb, cholesterol or albumin over the trial period.

Table 4 demonstrates the multivariable linear regression analysis for change in eGFR and uPCR at 24 months. After adjustment for baseline eGFR, age, ACEi and sex the change in eGFR with the addition of corticosteroids from baseline to 24 months trended towards significance (β coefficient: 18.63 ml/min/1.73 m², 95% CI [− 2.26–39.52]; (p = 0.078) (Table 4). There was no association observed for change in uPCR (β coefficient: 0.09 g/mmol, 95% CI -0.19-0.33); (p = 0.525) over the study period.

Table 4

Multivariable linear regression analysis for Change in eGFR and uPCR at 24 months

	Δ in eGFR (ml/min/1.73m2) at 24 months		Δ in UPCR (gm/mmol) (at 24 months
	β Coefficient (95%CI)	P = value	β Coefficient (95% CI)	P = value
On steroids	18.63 [−2.26–39.52]	0.078	0.09 [−0.19–0.37]	0.525
Baseline eGFR	−0.16 [− 0.47–0.16]	0.317	0.00 [− 0.00–0.01]	0.695
Age	0.37 [− 0.59–1.33]	0.435	0.01 [0.00–0.03]	0.031
ACE inhibition	−2.63 [− 23.48–18.21]	0.798	− 0.08 [− 0.36–0.20]	0.574
Sex, female	−9.67 [− 29.06–9.71]	0.316	0.11 [− 0.15–0.37]	0.384

Estimated glomerular filtration rate [eGFR], urine protein creatinine ratio [uPCR], angiotensin converting enzyme [ACE], confidence interval [CI]

Our study is the first open labelled prospective trial in Africa to assess the effect of corticosteroids on kidney function in patients with HIVAN treated at a single centre in Cape Town, South Africa. Important observations from this study include the following: (i) a significant increase in eGFR in patients treated with corticosteroids (without a significant reduction in proteinuria at last follow up) (ii) increased adverse events including risk of infections and all-cause mortality in the group treated with ART and adjuvant corticosteroids, and (iii) reduced interstitial inflammation seen on repeat biopsy in both arms without a significant difference between the groups.

Open image in new window

Previous studies have demonstrated a positive effect of corticosteroids on kidney function in patients with biopsy proven HIVAN [22, 23, 24]. In a study of 20 patients with HIVAN on dual ART therapy with oral prednisone, Smith et al. reported a reduction of serum creatinine as well as a significant reduction of proteinuria [22]. However, relapses were reported when prednisone was tapered and six patients developed serious infections on prednisone, with 11 reported deaths [22]. In another study of 21 patients, 12 of which were treated with prednisone, Eustace et al reported an 80% reduction in risk of ESRD, but a higher risk of infections in the treatment group [23]. Only one retrospective study of 19 patients has assessed the use of corticosteroids in the triple therapy era of ART. The 5 patients who received prednisone experienced an increase in creatinine clearance of 5.57 ml/min/1.73 m² per month (p = 0.003) [24].

The rational for the corticosteroid regimen used in our trial was based on the dosing schedule used to treat drug induced acute interstitial nephritis [AIN] as well as a regimen similar to that used by Smith et al. for steroid use in HIVAN [22, 26]. The tapering period was lengthened to a gradual taper over 6 months as the average length of time for HIV viral suppression after initiation of ART is 3–6 months [27, 28, 29]. The working hypothesis was that improving interstitial inflammation would improve renal function. Since HIV is the cause of interstitial inflammation in HIVAN it was important to allow enough time for the HIV to be controlled prior to cessation of the steroids.

This study demonstrates a positive improvement in eGFR in those receiving [ART + C] compared with those on [ART Alone] [Δ = 25mls/min (IQR: 15–51) vs 9 mls/min (IQR: 0–24), p = 0.008]. However, when the multivariable linear regression model was adjusted for baseline eGFR, although the change in eGFR over the study period was no longer significant the β coefficient [18.63] still demonstrated an improvement compared to ART alone. The lack of statistical significance is likely due to small numbers and the marked discrepancy of baseline eGFR between the two cohorts. Although proteinuria improved there was no statistical significant difference between the groups. On interval analysis, inflammation improved in both groups. Interstitial inflammation is a prominent histopathologic component of HIVAN [9, 11]. Inflammation may lead to interstitial fibrosis which is strongly correlated with poor renal prognosis in many forms of CKD including HIVAN [11, 30, 31]. .The lack of additional improvement of corticosteroids on proteinuria while improving eGFR may point to a benefit in dampening down the interstitial inflammatory response. The renal biopsy may have been performed too early and the sample size too small to see a significant response to those treated with corticosteroids.

There were 10 cases of TB identified prior to trial initiation. There were no further cases documented throughout the trial. Those with TB continued and completed their course of treatment. There were adverse events documented in the trial. There were 2 cases of herpes zoster treated without sequelae, 2 weeks after commencing ART and corticosteroids. There were deaths occurring in the early trial period in both groups with more deaths occurring in the corticosteroid arm. This could not be explained by differences in CD4 count or viral load suppression between the groups. [See Additional file 1 Table S1] This raises the question of a significant contributory effect of corticosteroids to sepsis in these patients.

However in contrast to our findings, a study conducted in South Africa using corticosteroids for the adjuvant treatment of tuberculous pericarditis used substantially higher doses of steroids [32]. This study had high numbers of HIV-infected patients and demonstrated a significant reduction in the incidence of constrictive pericarditis with no increased risk of mortality in those patients receiving corticosteroids. The trial did however demonstrate an increase in HIV associated cancers a finding not observed in our study [32]. The results of our trial suggest the need to reconsider the dose and duration of corticosteroid therapy. However, given that renal replacement therapy is rationed or limited in many regions of Africa the positive effect seen on kidney function needs to be balanced against the likelihood of sepsis and mortality.

There were several limitations of the study. Firstly, the sample size was small and the baseline eGFR was significantly lower in the corticosteroid arm. The randomization process used was via sealed envelope, although not ideal, was strictly adhered to. This highlights the need for a larger study defining hard endpoints of eGFR and proteinuria improvements with stratification into eGFR subgroups, to improve the validity of the results. Secondly patients and clinicians were not blinded to those who received corticosteroids, however the reason for this was additional opportunistic prophylaxis was required for this group. Thirdly at the start of this study there was no consensus on the histological definition of HIVAN. Fourthly, the length of time of follow up for the cohort was 24 months. For meaningful renal survival data, a longer follow up time period would be necessary. Lastly, pill counts were not used to assess compliance on corticosteroids.

Despite the observed improvement in eGFR in patients treated with corticosteroids, our study cannot currently justify the use of corticosteroids at 1 mg/kg with tapering over 6 months to treat patients with biopsy proven HIVAN due to the increased mortality seen. Given the lack of effect on proteinuria it is postulated that the positive effect on eGFR may be having an impact on the interstitial component seen in HIVAN. Future longitudinal studies need to be directed to understanding those patients most likely to benefit from corticosteroid therapy while identifying those at higher risk for septic events. A lower corticosteroid dose, shorter duration of therapy and closer follow up may improve outcomes in this patient group.