Heart muscle disease related to HIV infection: prognostic implications

Introduction

The cardiac complications of HIV infection and AIDS are now well documented.^{1 2} They include pericardial effusions, disturbances of rhythm, malignant infiltration, marantic endocarditis, and heart muscle disease. With improved clinical surveillance and treatment, more patients are surviving potentially fatal opportunistic infections only to succumb to neoplasia or end organ damage.3 Heart muscle disease is one such complication and seems destined to become an important cause of cardiac failure worldwide. The natural course of heart muscle disease that is related to HIV infection has not yet been established, largely because most published work has been based on small or cross sectional surveys. Anecdotal evidence exists that patients with pronounced impairment of left ventricular function have a poor prognosis, but this has yet to be confirmed by a formal survival study.4 5 6

A cohort of patients with HIV infection has been studied prospectively in Edinburgh since 1990. Most of the patients attend a single centre, where detailed records are kept. We investigated the natural course of the various forms of heart muscle disease and determined the effect of each condition on survival.

Subjects and methods

Over four years, 296 subjects with HIV infection (mean age 32.7 years (range 21.5 to 67.6)) from all the major groups at risk of HIV infection in Britain (203 injecting drug users, 52 homosexuals, 28 heterosexuals, seven bisexuals, three recipients of blood products, and three patients with multiple risk factors) were assessed with serial echocardiography, chest radiography, and electrocardiography. (One hundred and fifty four patients were reported on previously.2) According to criteria from the Centers for Disease Control,7 AIDS was diagnosed in 100 subjects, 89 subjects were classed as group IV without AIDS and 45 as group III, and 10 subjects had asymptomatic HIV infection. A contemporary classification was not available for 52 subjects.

ECHOCARDIOGRAPHY

We used phantom calibrated ultrasound machines (Sonos 100 and 1000, Hewlett Packard) with 2.5 MHz and 3.5 MHz transducers with subjects in the left lateral position. We obtained left ventricular M mode tracings from long and short axis parasternal views at the level of the mitral valve papillary muscles, and we used on screen callipers to measure end diastolic and systolic dimensions. We calculated the left ventricular fractional shortening as the difference between these two measurements divided by the end diastolic size. The echocardiograms were stored on video tape and were subsequently analysed by the operator and two independent observers blinded to the clinical history of the subjects. M mode measurements were unobtainable in 20 subjects, but cardiac assessment based solely on two dimensional imaging was normal in all these cases.

Patients found to have cardiac dysfunction were grouped into three categories: dilated cardiomyopathy, isolated right ventricular dysfunction, and borderline left ventricular dysfunction (box). None of the subjects with cardiac dysfunction had received anthracycline drugs during the study.

We obtained CD4 cell counts by lymphocyte immunophenotyping using a flow cytometer (FACScan, Becton Dickinson). We obtained information on the clinical status of each subject and, where appropriate, the cause and date of death from hospital records and postmortem reports.

STATISTICAL METHODS

The primary end point was mortality from a condition related to AIDS, and survival times were censored at death from other causes. Survival curves were obtained with the Kaplan-Meier estimate, and Cox's proportional hazards regression (with BMDP software) was used to compare the three groups with heart muscle disease with the rest of the cohort, who had structurally normal hearts.

The association between CD4 cell counts and cardiac complications was evaluated with a KruskalWallis test to compare the group with normal hearts with the groups with left or right ventricular dysfunction. A Mann-Whitney test was used to compare these three groups with the group with cardiomyopathy.

Results

FIG

Top: Survival curves for 296 patients who are HIV positive with structurally normal hearts or cardiac dysfunction. Bottom: Survival time to death related to AIDS in 81 subjects with CD4 cell count <20 × 10⁶/l

• Download figure
• Open in new tab
• Download powerpoint

The mean CD4 cell count of the subjects was 153 × 10⁶/l (range 0-1178) (normal range 500-1400). Over 90% (268/296) of subjects showed evidence of immunosuppression, with a CD4 count of < 400 × 10⁶/l, of whom 146 (54%) had a count of <100 × 10⁶/l.

Echocardiographic evidence of cardiac dysfunction was found in 44 subjects (dilated cardiomyopathy in 13, right ventricular dilatation in 12, and borderline left ventricular dysfunction in 19). Chest radiography was less useful, with cardiomegaly (cardiothoracic ratio of >0.5) present in only 46% (6/13) of the patients with dilated cardiomyopathy and in only one patient with isolated left ventricular dysfunction. The electrocardiographic results were unremarkable other than showing non-specific T wave changes, which occurred in all groups.

Symptoms attributable to, and clinical signs of, heart failure were common in the subjects with dilated cardiomyopathy (9/13 (69%)) but were less so in those with borderline left ventricular dysfunction (7/19 (37%)). Breathlessness was common in the subjects with right ventricular dysfunction (6/12 (50%)) but was often attributable to pulmonary disorders.

In all, 123/296 subjects died during the study. Of these, 99 died of conditions related to AIDS and the rest of other causes, including hepatitis (nine) and drug overdose (six). Among those who died of conditions related to AIDS were 12 of the 13 subjects (92%) with dilated cardiomyopathy, 5 of the 12 (42%) with right ventricular dysfunction, and 8 of the 19 (42%) with borderline left ventricular function.

CD4 cell counts were significantly lower (P < 0.001) in the subjects with dilated cardiomyopathy than in the subjects in the three other groups, but no significant difference (P = 0.27) occurred between the counts for subjects with isolated right or borderline left ventricular dysfunction and those for the subjects with normal hearts (table). As the patients with dilated cardiomyopathy had more advanced HIV disease (disease lasting from seroconversion to death) Cox's regression was used to correct for the CD4 cell count (and also for age, sex, status according to the criteria from the Centers for Disease Control, interval between echocardiography and the CD4 cell count, and risk group). The figure (top) shows survival curves for deaths related to AIDS for each group. Cox's analysis showed that the subjects with cardiomyopathy had a significantly worse outlook than the patients with normal hearts (χ² = 23.43, P < 0.001; hazard ratio 11.68 (95% confidence interval 4.32 to 31.58)). The subjects with isolated right and borderline left ventricular dysfunction (with hazard ratios 1.17 (0.51 to 2.70) and 1.48 (0.56 to 3.95) respectively) did not differ significantly from those with structurally normal hearts.

The excess mortality in subjects with cardiomyopathy remained significant at P < 0.001 even when analysis was restricted to the subjects with CD4 cell counts of < 20 × 10⁶/l (figure (bottom)) or when all cause mortality was assessed. Median survival to death related to AIDS was 101 days (95% confidence interval 42 to 146) in those with cardiomyopathy compared with 472 (383 to 560) in 59 of the subjects with normal hearts. Furthermore, the mortality of the 13 subjects with cardiomyopathy was significantly increased when compared with 13 of the patients with normal hearts (median survival 407 days) who were individually matched for age, sex, risk factor for HIV infection, and CD4 cell count (P < 0.01).

Discussion

This study confirms previous work that indicated that the prevalence of heart muscle disease in patients with HIV infection is about 15%. Heart muscle disease that is related to HIV infection takes three forms: global left ventricular dysfunction (particularly in late stage HIV disease), isolated right ventricular dilatation, and borderline left ventricular dysfunction. Cardiac dysfunction is rare in control patients with haematological malignancy6 and in drug users without HIV infection;8 it is therefore probably due to a direct effect of HIV infection rather than an effect of the cachexia in patients with AIDS. Little information exists about the natural course of these conditions, although anecdotal reports indicate that dilated cardiomyopathy is associated with a poor prognosis.4 5 6

We have shown that patients with dilated cardiomyopathy generally die early from conditions related to AIDS. This appears to be independent of these patients' CD4 cell count at presentation. Although the statistical power to detect a moderately higher risk is low, the outlook for patients with other forms of heart muscle disease is much less gloomy and may reflect the potentially reversible nature of these conditions. Left ventricular dysfunction may be caused by a self limiting myocarditis that does not progress to overt left ventricular failure. Isolated right ventricular dysfunction may be transient and is possibly related to the pulmonary hypertension associated with parenteral drug use or recurrent bronchopulmonary infection9 rather than to primary myocardial disease.

Left ventricular failure has rarely been reported as a cause of death in this series, even in patients with cardiomyopathy. We suspect that it was a contributory factor in many cases but may have been overlooked as the features of heart failure are often mistakenly attributed to opportunistic pulmonary infections or anaemia, which are common in patients with HIV infection.^{2 6} Similarly, cardiac complications tend to be underdiagnosed even in the face of convincing postmortem evidence.10

This study shows that dilated cardiomyopathy is an independent adverse prognostic factor in patients with HIV infection and that it is strongly associated with a very low CD4 cell count. Other conditions with prognostic significance include oral candidiasis,^{11 12} hairy leucoplakia,^{11 13} severe herpes zoster,14 and constitutional symptoms,11 which are all associated with an increased risk of progression from asymptomatic HIV infection to AIDS.

The results of the Concorde trial indicated that it is inappropriate to rely solely on the CD4 cell count when assessing the need for antiretroviral treatment or the long term prognosis of patients with HIV infection.15 A decline in CD4 cell count correlates well with disease progression in population studies,11 16 17 but a wide intraindividual variation can exist in both the absolute count and the rate of cell loss over time, reducing the predictive power of this value for individual patients. Decisions on treatment should therefore take into account both the trend of serological results and the clinical factors (including the presence of heart muscle disease).18

The cause of heart muscle disease related to HIV infection remains unknown, but many cases seem to be related to an idiopathic lymphocytic myocarditis that is a common postmortem finding in patients with ventricular dysfunction.1 This may be due to a direct or indirect effect of HIV infection or more rarely, cytomegalovirus infection,19 but other potential pathogenic factors include nutritional deficiencies, opportunistic infections, and a cardiotoxic effect of antiretroviral drugs.20 Endomyocardial biopsies and appropriate serological tests for cytomegalovirus or Toxoplasma gondii, however, have shown no evidence of opportunistic infection in our patients.2

The importance of cardiac involvement in HIV infection should not be ignored. The World Health Organisation has estimated that 38-110 million people worldwide will be HIV positive by the turn of the century. Even if only 4% of these develop dilated cardiomyopathy, HIV will constitute an important cause of heart failure. The outlook for patients with dilated cardiomyopathy is clearly poor and seems to be independent of all other potential prognostic factors. Clinical trials should be set up to evaluate the efficacy of conventional treatment of heart failure in patients with heart muscle disease related to HIV infection.

We thank Mrs Jean Cunningham for her secretarial help in preparing this manuscript. PFC is funded by the British Heart Foundation.