Bone marrow morphology in human immunodeficiency virus-infected South Africans with and without tuberculosis


Although the morphological bone marrow changes associated with human immunodeficiency virus (HIV) infection have been well characterized (Zon & Groopman, 1988; Karcher & Frost, 1991; Bain, 1997), most studies have come from industrialized countries and co-existent infections have sometimes confounded interpretation of these results. This study was undertaken to determine the morphology of bone marrow aspirates in HIV-infected patients with and without tuberculosis involving the bone marrow in a South African population.

Research was carried out on 69 HIV-infected patients from Chris Hani Baragwanath Hospital, Soweto, South Africa, who had undergone bone marrow examinations (for cytopenia, unexplained fever or wasting) between 1997 and 1998 and who had a CD4 count performed within 3 months of the investigation. Detailed cytological examinations were performed by two assessors who were blinded to clinical details. Tuberculosis of the bone marrow was diagnosed in 31 subjects on the basis of acid-fast bacilli, positive culture for Mycobacterium tuberculosis of marrow aspirate material, or granulomas noted on trephine biopsy examination, either with caseation necrosis or with proof of tuberculosis elsewhere. No subject had infections other than tuberculosis involving the bone marrow, granulomas of unknown cause or malignancies.

There were 31 men and 38 women aged 16–65 years. Granulomas were identified in 25 (81%) bone marrow trephine biopsies of patients with tuberculosis of the bone marrow. No patients were receiving anti-retroviral therapy, five were on cotrimoxazole, and three were folate-deficient, of whom one was vitamin B12-depleted. Peripheral blood cytopenia was found in 61 (88%) patients. The morphological findings of the bone marrow aspirates are shown in Table I. There were no significant differences between the groups with and without tuberculosis. Severe dysplasia was a prominent finding in all three cell lines. In the group without tuberculosis, increased erythropoiesis in patients with CD4 cell counts of 0–100 × 106/l was significantly less common than in patients with counts of 101–500 × 106/l (14% vs. 50%; P = 0·03). In the group with tuberculosis, there was significantly more hypercellularity (92% vs. 55%; P = 0·048) and less increased granulopoiesis (17% vs. 62%; P = 0·01) in subjects with CD4 cell counts of 0–100 × 106/l than in patients with counts of 101–500 × 106/l. Sixteen (23%) patients had haemophagocytic cells involving at least 2% of every 500 nucleated marrow cells, 16% of the HIV-only group and 32% of the tuberculosis group (P = 0·11).

Table I.  The frequency of peripheral blood cytopenia and bone marrow morphological abnormalities in HIV-infected patients with and without bone marrow tuberculosis.

HIV without tuberculosis (n = 38)HIV with tuberculosis (n = 31)
Frequency (%)Frequency (%)
Increased myeloid:erythroid ratio1816
Decreased myeloid:erythroid ratio2932
Increased granulopoiesis4735
Decreased granulopoiesis1113
Increased erythropoiesis3729
Decreased erythropoiesis1116
Increased megakaryopoiesis2722
Increased iron stores8484
Increased macrophages3958

HIV-related abnormalities in bone marrow morphology overshadowed any alterations in marrow cytology associated with tuberculosis, as shown by the lack of significant difference in morphology between the two groups. The marrow morphological changes associated with tuberculosis are non-specific and commonly include increased iron stores and macrophage numbers often with haemophagocytosis, while granulomas are associated with only minor non-specific alterations in marrow morphology (Bain et al, 1996).

The frequencies of the marrow morphological abnormalities in this population were higher than expected (Zon & Groopman, 1988; Karcher & Frost, 1991; Bain, 1997) with respect to hypercellularity (usually 50–60% of cases) and dysplasia. Dysgranulopoiesis is usually seen in 70% of patients, erythrocytic dysplasia in 50–60% of cases and dysplasia of megakaryocytes in one third of patients. The overall pattern of marrow changes was similar to that described in industrialized countries (Zon & Groopman, 1988; Karcher & Frost, 1991; Bain, 1997).

Although morphological changes of the bone marrow are said to be increased in more advanced HIV infection, a CD4 cell count < 101 × 106/l was not a good predictor of a difference with only three non-specific changes being demonstrated. This lack of discrimination has been found for CD4 counts < 400 × 106/l (Karcher & Frost, 1991) and for clinical staging (Adewuyi et al, 1999).

In conclusion, we have demonstrated profound HIV-related haematological changes in this African population that occur early in the course of HIV infection and are unaffected by concomitant tuberculosis involving the bone marrow.