Changing clinical profile and factors associated with liver enzyme abnormalities among HIV-infected persons

Abstract

The spectrum of liver disease among HIV-infected patients is changing. In the era of antiretroviral therapy, opportunistic infections are diminishing and deranged liver function appears to be due usually to drug-induced liver injury, alcohol, non-alcoholic steatohepatitis (NASH) or chronic hepatitis B. To test this hypothesis, 98 HIV-positive patients with deranged liver function were compared with matched HIV-positive patients with normal liver function and likewise matched HIV-negative patients with normal liver function tests.

Keywords

AIDS Asia epidemiology digestive

Introduction

Hepatobiliary diseases make a major contribution to the morbidity and mortality of HIV-infected people.¹ With the introduction of antiretroviral treatment (ART), HIV-infected patients are experiencing longer life expectancy and some reports suggest liver disease as the leading cause of death.² In these patients, hepatic abnormalities are attributed to chronic viral hepatitis (B and C) as co-infections, presumably owing to shared routes of transmission.³ Opportunistic pathogens (viz. cytomegalovirus) and tumours (viz. lymphoma and Kaposi’s sarcoma) were common causes of hepatobiliary derangement in the past, but these are diminishing.⁴ Tuberculosis is, however, increasingly prevalent in low- and middle-income countries (LMICs) in HIV-infected patients and this may itself result in liver function derangement. Elevation of hepatic enzymes may also be the result of multiple factors such as chronic alcohol consumption, drug addiction, lipid-lowering drugs, ART itself or hereditary diseases.⁵ In addition, it has been proposed that HIV itself causes direct damage to hepatocytes.⁶ Furthermore, with a rising incidence of obesity, reports have suggested that non-alcoholic fatty liver disease (NAFLD) may be important both in the general population and in HIV-infected patients.^7,8 Information from the latter countries is normally based either on liver histology or clinical diagnosis.⁹ The study was therefore carried out incorporating both clinical and histological parameters for the evaluation of deranged liver function in HIV-infected patients.

Methods and Materials

This prospective cohort study was conducted at a tertiary care centre in Mumbai. The study population consisted of confirmed HIV-infected patients with abnormal liver biochemical tests attending from December 2013 to June 2014. Patients with abnormal liver enzymes (levels >1.5 the normal range), either symptomatic or asymptomatic, taking ART or not, were included. HIV-positive patients aged below 18 years and pregnant women were excluded. Equal numbers of age- and sex-matched HIV-positive and HIV-negative patients with normal liver function as controls were taken from the general outpatient clinic (Figure 1). All participants provided written informed consent, and the study was approved by the local Institutional Ethics Committee.

Figure 1.

Flow diagram of study.

Demographic details of all patients were recorded. Clinical history included presenting complaints of recent illness, duration of HIV-positivity, details of alcohol ingestion, duration of ART, anti-tubercular, diabetic, anti-hypertensive and lipid-lowering medication were recorded. Alcoholic liver disease was excluded as cause of disease if intake was <30 g/day and <150 g/week for men or <20 g/day and <100 g/week for women.^10,11 A detailed clinical examination was performed in all patients. The body mass index (BMI) was calculated as kg/m². A recent CD4 count, the type of antiretroviral drugs used and the temporal relationship between symptoms and any medication received was recorded. Drug-induced liver injury (DILI) was diagnosed based on such temporal relationship and depending on resolution of liver function derangement after stopping the culprit drug and exclusion of other causes of liver injury.¹² All patients had full blood count (FBC), fasting blood sugar, levels of serum transaminases (ALT and AST), bilirubin, alkaline phosphatase (ALP), total protein, albumin, prothrombin time (PT), international normalized ratio (INR), fasting serum total cholesterol, serum triglyceride, hepatitis B surface antigen (HBsAg), hepatitis C antibody (anti-HCV) and renal function tested. Likewise, all had abdominal ultrasonography of the hepatobillary system performed. In selected patients, hepatitis B core antibody (total anti HBc), anti-nuclear antibody, anti-smooth muscle antibody, anti-liver-kidney microsome antibodies, serum cerulopalsmin and 24 h urinary copper were also tested. In patients with HBsAg or anti Hbc positive status, tests for HBV DNA and HBeAg were performed. Liver biopsy was carried out when the cause of deranged liver function was not identified after virological, radiological and immunological work-up. The AST/platelet ratio index (APRI) was calculated using the formula AST level/reference value/platelet count (×10⁹/L) × 100.

Statistical analysis

The Mann–Whitney test was used for comparison between the two main HIV-positive groups. One-way ANOVA was used in analysis of quantitative data between all three groups if data passed a ‘Normality test’ and by the Kruskal–Wallis test if data failed. Associations among variables in patients who died compared with those who improved were defined with the help of Binary Logistic Regression. A P value <0.05 was taken as significant. Statistical analysis was done using SPSS Version 16.0 for Windows.¹³

Results

A total of 135 patients were assessed for eligibility, of whom 98 met the inclusion and exclusion criteria for our study. Equal numbers of age- and sex-matched HIV-positive and HIV-negative patients with normal liver function were taken from our general outpatient clinics (Figure 1). The comparison of various parameters among the study groups is shown in Table 1. The alcohol consumption in all three groups was not significantly different (P = 0.472). A longer duration of HIV-positivity and low CD4 count was significantly associated with deranged liver function (P < 0.05) but the duration of ART was not found to be associated (Table 2). Of those patients on ART, LFTs were abnormal in 63 (64.3%) and normal in 59 (60.2%). Of the 63 patients on ART with abnormal LFTs, 22 (34.9%) were taking tenofovir, lamivudine and nevirapine (TLN), 20 (31.7%) were taking tenofovir, lamivudine and efavirenz (TLE), 16 (25.4%) were taking zidovudine, lamivudine and nevirapine (ZLN) and five (7.9%) were taking tenofovir, lamivudine, atazanavir and ritonavir (TLAR). The incidence of deranged liver function was not statistically different among the various ART regimens (Table 3).

Table 1.

Comparison of various parameters among three study groups.

Characteristics	Group	Median	IQR	Analysis
Age (years)	Deranged LFT	42	13	Kruskal–Wallis test
	Normal LFT	40	9.25	H value	P value
	Non-HIV	40	15.25	0.713	0.700
Serum cholesterol (mg/dl)	Deranged LFT	165	42	Kruskal–Wallis test
	Normal LFT	145	49.25	H value	P value
	Non-HIV	143	51	15.256	<0.001
Serum triglyceride (mg/dl)	Deranged LFT	150	69.5	Kruskal–Wallis test
	Normal LFT	136.5	51.25	H value	P value
	Non-HIV	121	33.75	26.247	<0.001
Haemoglobin (g/dl)	Deranged LFT	11.3	3.55	Kruskal–Wallis test
	Normal LFT	11.8	2.8	H value	P value
	Non-HIV	12.4	1.8	47.486	<0.001
		Mean	Std. Dev
WBC (10⁹/L)	Deranged LFT	6837.55	2826.60	One-way ANOVA test
	Normal LFT	6971.53	2934.23	F value	P value
	Non-HIV	6090.92	1954.76	3.242	0.041
BMI (kg/m²)	Deranged LFT	20.47	2.94	One-way ANOVA test
	Normal LFT	21.4	3.33	F value	P value
	Non-HIV	23.53	3.03	25	<0.001
Platelet count (10⁹/L)	Deranged LFT	232.94	183.77	One-way ANOVA test
	Normal LFT	228.28	98.87	F value	P value
	Non-HIV	255.07	103.65	1.11	0.331

BMI, body mass index; WBC, white blood count.

Table 2.

Comparison of various parameters in HIV patients with deranged LFT and normal LFT.

Characteristic	Group	Median	IQR	Mann–Whitney test	P value
Duration of HIV (months)	Deranged LFT	36.00	60.50	–2.407	0.016
Duration of HIV (months)	Normal LFT	24.00	24.00	–2.407	0.016
CD4 count (cells/mL)	Deranged LFT	267.00	232.50	–2.866	0.004
CD4 count (cells/mL)	Normal LFT	328.00	225.50	–2.866	0.004
Duration of ART (months)	Deranged LFT	24.00	46.00	–1.874	0.061
Duration of ART (months)	Normal LFT	12.00	30.00	–1.874	0.061
Serum cholesterol (mg/dl)	Deranged LFT	165.00	42.00	–2.786	0.005
Serum cholesterol (mg/dl)	Normal LFT	145.00	49.25	–2.786	0.005
Serum triglyceride (mg/dl)	Deranged LFT	150.00	69.50	–2.407	0.016
Serum triglyceride (mg/dl)	Normal LFT	136.50	51.25	–2.407	0.016

Table 3.

Distribution of lipid profile and liver enzyme as per ART regimen.

ART regimen	TLN	TLE	ZLN	TLAR	F value	P value
Abnormal LFT (Group 1)	22	20	16	5	N/A	N/A
Normal LFT (Group 2)	25	16	16	2	N/A	N/A
Serum cholesterol (mg/dl) (Group 1)	170.36 ± 44.95	175.60 ± 74.61	210.44 ± 80.38	186.60 ± 74.60	1.229	0.307
Serum triglyceride (mg/dl) (Group 1)	173.05 ± 49.96	162.40 ± 56.66	155.00 ± 51.94	147.20 ± 61.73	0.523	0.668
AST (U/L) (Group 1)	104.45 ± 75.70	188.05 ± 171.74	279.81 ± 524.55	75.20 ± 37.78	1.382	0.257
ALT (U/L) (Group 1)	94.45 ± 49.49	139.75 ± 188.38	445.00 ± 137.06	54.20 ± 38.64	1.612	0.196
APRI (Group 1)	2.18 ± 3.32	2.17 ± 1.89	3.57 ± 4.81	0.93 ± 1.00	1.050	0.377

ALT, alanine aminotransferase; APRI, AST platelet ratio index; AST, aspartate aminotransferase; N/A, not applicable; TLAR, Tenofovir (TDF) + Lamivudine (3TC) + Atazanavir (ATV) + Ritonavir (RTV); TLE, Tenofovir (TDF) + Lamivudine (3TC) + Efavirenz (EFV); TLN, Tenofovir (TDF) + Lamivudine (3TC) + Nevirapine (NVP); ZLN, Zidovudine (AZT/ZDV) + Lamivudine (3TC) + Nevirapine (NVP).

The most common causes of liver dysfunction in decreasing order were: DILI, either anti-tubercular (ATT) or ART (29%), chronic hepatitis B (18%), alcoholic liver disease (11%) and NASH (7%). Other causes were rare. Among the 98 patients with deranged liver function, 26 were HBsAg-positive, nine were hepatitis B e antigen (HBeAg) positive and 23 patients had detectable HBV DNA (mean 2.1 × 10⁷). Of the 26 HBsAg-positive patients, five had other factors present such as excessive alcohol consumption and drugs as contributing factors. Contrary to reports from elsewhere, hepatitis C was seen in only two patients. Among 16 patients who underwent liver biopsy, seven had normal liver histology, five had steatohepatitis, two had granuloma and two were found to have cholestasis. The cause of liver disease was unexplained in 12 patients, of whom seven had normal liver histology and five recovered to normal liver function during their evaluation. Hepatomegaly was found on ultrasonography in 50% of patients who underwent liver biopsy. During the mean 6-month follow-up period, eight patients (8.2%) died. Logistic regression analysis for outcome showed that age, CD4 count, total serum bilirubin, AST, ALT, APRI, serum protein, serum creatinine, haemoglobin and white blood cell count were not associated with improvement.

Discussion

Liver diseases in HIV-infected individuals is an emerging issue and there is heterogeneity in its aetiology in different parts of the world. The aetiology of liver diseases is unexplained in nearly 50% of asymptomatic patients in some recent studies. Liver diseases accounts for 10–18% of observed deaths and ranks as one of the first cause of death in HIV patients.¹⁴ Heterosexual transmission accounts for 84.7% in LMICs such as India.¹⁵ With increasing industrialization, injecting drug use is a major a risk factor for both HIV transmission as well as associated HBV/HCV co-infection. It also accounts for the higher prevalence of alcohol abuse and obesity leading to increased liver related death. A study from California, found that in HIV-infected persons, NAFLD was the most common cause of abnormal liver function (30%) followed by excessive alcohol use (13%), chronic hepatitis B (9%), chronic hepatitis C (5%) and more than one diagnosis in 10% of patients. Eighteen percent of patients were obese with a mean BMI of 25.8 kg/m². However, despite having no evidence of obesity, we found a considerable prevalence of NASH (7%) among our HIV patients which mimics lean NASH seen in Asian countries.¹⁶ In HIV–TB co-infected patients, evaluation of the cause of abnormal liver function is very challenging as it is difficult to differentiate between DILI and tuberculosis immune reconstitution inflammatory syndrome (TB-IRIS). Similar to DILI, TB-IRIS patients can present with jaundice after initiating ART. Histopathological findings which suggest TB-IRIS in liver tissue include large epithelioid granulomata which are again difficult to distinguish from tuberculous granulomata. A study from Tanzania reported that the incidence of DILI in patients on concomitant HIV and TB therapy was almost twice (i.e.10%) that of those treated with efavirenz-based ART only (5.9%).¹⁷ The most common cause of deranged LFTs in our study was DILI (16% in ATT and 14% in ART). There was no statistically significant difference in the incidence of DILI among the various ART regimens although some antiretroviral drugs have propensity to cause DILI more frequently than others. Nucleoside agents may cause direct hepatotoxicity and steatosis due to inhibition of mitochondrial DNA polymerase-γ.¹⁸

HIV-infected patients have been reported to exhibit an abnormal lipid profile which has correlation with CD4 counts.¹⁹ Most studies have showed that triglyceride is increased in HIV patients compared to the general population. This increase is due to raised very low density lipoprotein levels (VLDL) and alteration in cytokine like interferon, interleukins and TNF.²⁰ The level of serum cholesterol has conflicting results in HIV patients. Generally cholesterol will decrease in advanced HIV disease and its level will increase after HAART treatment.²¹ We found significantly high serum cholesterol and triglyceride in the deranged LFT group compared to the normal LFT group; this could explain the increased incidence of NASH in HIV-infected patients. Liver test abnormalities among HIV patients receiving protease inhibitors may be largely related to an increased lipid level, hence adequate control of lipid levels may be useful in preventing liver enzyme elevations and perhaps ongoing liver dysfunction among this patient population.

Haematological abnormalities are prevalent in HIV-infected patients which can cause hyperbilirubinemia (mainly indirect) and elevation of serum AST. Low haemoglobin in the HIV population was associated with a higher death rate in both the low CD4 and high CD4 groups and anaemia was independently associated with a higher mortality at 1 year of follow-up.²² Similar findings in our patients indicate that as the immunosuppression worsens, the probability of illness due to liver disease increases.²³

The APRI >1.5 is a non-invasive marker of significant liver fibrosis in patients with HIV/HCV co-infection;²⁴ 47% of our patients had APRI >1.5 and seven patients out of eight who died had APRI >1.5. Significant liver fibrosis by APRI score has been reported in 8% of HIV-mono-infected patients.²⁵ Diabetes mellitus and high HIV viral load were significant risk factors for fibrosis. While HBV/HCV co-infections are obvious causes of liver fibrosis, NAFLD and NASH are other emerging causes leading to chronic liver disease.

Liver test abnormalities may be transient in nature in up to 30% of the general population.²⁶ Liver histology is critical in the evaluation of abnormal liver tests, especially if the abnormality persists even after eliminating the offending aetiological factor. Initial studies on liver histology in HIV-infected patients revealed that infectious diseases were the most common cause of liver test abnormalities in LMICs.²⁷ There was a low prevalence of opportunistic infection in our study as most of the patients were on ART with an average CD4 count >310/mm³ and with good performance status. Consistent with a recent study from Africa, our results indicate that DILI due to ART is the most common cause of abnormal liver enzymes.²⁸ Two of our patients had HIV-induced papillary stenosis and one had nodular regenerative hyperplasia (NRH). The latter is a potential cause of non-cirrhotic portal hypertension in patients with HIV. It is being increasingly recognised among HIV-infected individuals as they experience increased longevity and longer exposure to medications.²⁹ Most NRH patients have normal liver enzymes and histopathological evaluation remains critical in excluding cirrhosis and other causes of portal hypertension.³⁰ Didanosine (DDI) has been strongly correlated with ART-associated NRH. Association with azathioprine (AZA), 6-mercaptopurine, 6-thioguanine, busulphan and cyclophosphamide has also been reported in recent studies.³¹ The pathogenesis of NRH is believed to be related to the differential blood supply to the liver with areas of ischemia and compensatory hypertrophy which leads to the formation of hepatic nodules.³²

During 6 months of follow-up, eight of our patients (8.2%) died. Even though most of the deaths were in patients with alcohol-related liver disease (alcoholic hepatitis and cirrhosis), we did not find any significant association between various parameters and mortality. In a study from Switzerland, 14.5% deaths in HIV patients were from liver-related causes and predictors of mortality were latest CD4 cell count, age, intravenous drug use, HCV infection and active HBV infection. There was no relationship with duration of ART.³⁴

A strength of our study is the extensive evaluation of each case to identify the potential causes of liver disease in an HIV-infected clinic population. Patient evaluation included a detailed history with records of medication use and questionnaire of alcohol use, physical examination, laboratory tests, abdominal ultrasound and liver histology. Very few studies from LMICs have extensively investigated HIV patients with deranged LFTs.³³ There are some limitations in this study. It is hospital-based and the sample size of study population is relatively low with a follow-up of only 6 months. Hepatitis B core antibody was done in selected patients and occult hepatitis B could have been missed in a few patients. However, this is unlikely to affect our results, because none of the patients in whom Hepatitis B core antibody was not done had any other evidence of chronic liver disease.

Conclusion

We have demonstrated that DILI is the most common cause of abnormal liver tests in HIV-infected patients. Serum cholesterol and triglyceride levels are significantly elevated in these patients contributing to the rising incidence of NASH. Liver test abnormalities may be transient in nature in the HIV population, thus performing biopsy in selected patients after careful clinical assessment provides valuable information in clinically complex patients. This approach is more realistic and cost-effective in resource-limited nations rather than an aggressive approach of performing a biopsy in every patient for an evaluation of abnormal liver function. Even after such an approach abnormal LFT was unexplained in around 12% of our patients.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Bica

McGovern

Dhar

. Increasing mortality due to end-stage liver disease in patients with human immunodeficiency virus infection. Clin Infect Dis 2001; 32: 492–497.

Weber

Sabin

Friis-Moller

. Liver-related deaths in persons infected with the human immunodeficiency virus: the D:A:D study. Arch Intern Med 2006; 166: 1632–1641.

Gupta

Singh

. Hepatitis B and C virus co-infections in human immunodeficiency virus positive North Indian patients. World J Gastroenterol 2006; 12: 6879–6883.

Pol

Lebray

Vallet-Pichard

. HIV infection and hepatic enzyme abnormalities: intricacies of the pathogenic mechanisms. Clin Infect Dis 2004; 38: S65–S72.

Battegay

. Choice of initial therapy. J Int AIDS Soc 2014; 17(Suppl. 3): 19488–19488.

Mata-Marín

Gaytán-Martínez

Grados-Chavarría

. Correlation between HIV viral load and aminotransferases as liver damage markers in HIV infected naive patients: a concordance cross sectional study. Virol J 2009; 6: 181–181.

Crum-Cianflone

Collins

Medina

. Prevalence and factors associated with liver test abnormalities among human immunodeficiency virus-infected persons. Clin Gastroenterol Hepatol 2010; 8: 183–191.

Vispo

Moreno

Maida

. Noncirrhotic portal hypertension in HIV-infected patients: unique clinical and pathological findings. AIDS 2010; 24: 1171–1176.

Ocama

Katwere

Piloya

. The spectrum of liver diseases in HIV infected individuals at an HIV treatment clinic in Kampala, Uganda. Afr Health Sci 2008; 8: 8–12.

10.

Mandayam

Jamal

Morgan

. Epidemiology of alcoholic liver disease. Semin Liver Dis 2004; 24: 217–232.

11.

Day

. Who gets alcoholic liver disease: nature or nurture? (extended abstract). Acta Gastroenterol Belg 2003; 66: 290–291; discussion 292–293.

12.

Chalasani

Hayashi

Bonkovsky

. Practice Parameters Committee of the American College of Gastroenterology. ACG Clinical Guideline: the diagnosis and management of idiosyncratic drug-induced liver injury. Am J Gastroenterol 2014; 109: 950–966.

13.

SPSS Inc. Released 2007. SPSS for Windows, Version 16.0. Chicago, IL: SPSS Inc.

14.

Ruppik M, Ledergerber B, Rickenbach M, et al. Changing patterns of causes of death: SHCS, 2005 to 2009. Paper presented at: 18th Conference on Retroviruses and Opportunistic Infections; February 27–March 2, 2011; Boston, MA.

15.

Ramesh

Gandhi

Rao

. Clinical profile of human immunodeficiency virus patients with opportunistic infections: A descriptive case series study. Int J Appl Basic Med Res 2015; 5: 119–123.

16.

Das

Chowdhury

. Lean NASH: distinctiveness and clinical implication. Hepatol Int 2013; 7(Suppl. 2): 806–813.

17.

Mugusi

Ngaimisi

Janabi

. Liver enzyme abnormalities and associated risk factors in HIV patients on efavirenz-based HAART with or without tuberculosis co-infection in Tanzania. PLoS One 2012; 7: e40180–e40180.

18.

Bruno

Sacchi

Filice

. Mitochondrial toxicity in HIV-HCV coinfection: It depends on the choice of antiretroviral drugs? Hepatology 2002; 35: 500–501.

19.

Pasupathi

Bakthavathsalam

Saravanan

. Changes in CD4 count, lipid profile and liver enzymes in HIV infection and AIDS patients. J Appl Biomed 2008; 6: 139–145.

20.

Rogowska-Szadkowska

Borzuchowska

. The level of triglycerides, total cholesterol and HDL cholesterol in various stages of human immunodeficiency virus (HIV) infection. Pol Arch Med Wewn 1999; 101: 145–150.

21.

National AIDS Control Organization (NACO) and Ministry of Health and Family Welfare. Natural History and Clinical Manifestation of HIV/AIDS. Specialist Training and Reference Module, New Delhi: Government of India, 2003.

22.

De Santis

Brunetta

Vilar

. Hematological abnormalities in HIV-infected patients. Int J Infect Dis 2011; 15: e808–811.

23.

Price

Seaberg

Badri

. HIV monoinfection is associated with increased aspartate aminotransferase-to-platelet ratio index, a surrogate marker for hepatic fibrosis. J Infect Dis 2012; 205: 1005–1013.

24.

Carvalho-Filho

Schiavon

Narciso-Schiavon

. Optimized cutoffs improve performance of the aspartate aminotransferase to platelet ratio index for predicting significant liver fibrosis in human immunodeficiency virus/hepatitis C virus co-infection. Liver Int 2008; 28: 486–493.

25.

DallaPiazza

Amorosa

Localio

. Prevalence and risk factors for significant liver fibrosis among HIV-monoinfected patients. BMC Infect Dis 2010; 10: 116–116.

26.

Lazo

Selvin

Clark

. Brief communication: clinical implications of short-term variability in liver function test results. Ann Intern Med 2008; 148: 348–352.

27.

Lanjewar

Rao

Kulkarni

. Hepatic pathology in AIDS: a pathological study from Mumbai, India. HIV Med 2004; 5: 253–257.

28.

Sonderup

Wainwright

Hall

. A clinicopathological cohort study of liver pathology in 301 patients with human immunodeficiency virus/acquired immune deficiency syndrome. Hepatology 2015; 61: 1721–1729.

29.

Sood

Castrejón

Saab

. Human immunodeficiency virus and nodular regenerative hyperplasia of liver: A systematic review. World J Hepatol 2014; 6: 55–63.

30.

Scourfield

Waters

Holmes

. Non-cirrhotic portal hypertension in HIV-infected individuals. Int J STD AIDS 2011; 22: 324–328.

31.

Hartleb

Gutkowski

Milkiewicz

. Nodular regenerative hyperplasia: evolving concepts on underdiagnosed cause of portal hypertension. World J Gastroenterol 2011; 17: 1400–1409.

32.

Shimamatsu

Wanless

. Role of ischemia in causing apoptosis, atrophy, and nodular hyperplasia in human liver. Hepatology 1997; 26: 343–350.

33.

Mulu

Gidey

Chernet

. Hepatotoxicity and associated risk factors in HIV-infected patients receiving antiretroviral therapy at Felege Hiwot Referral Hospital, Bahirdar, Ethiopia. Ethiop J Health Sci 2013; 23: 217–226.

34.

Smith

Sabin

Lundgren

. Factors associated with specific causes of death amongst HIV-positive individuals in the D:A:D Study. AIDS 2010; 24: 1537–1548.