Most of the 40 million people living with HIV/AIDS at the end of 2003 were in low- and middle-income countries, where fewer than 5% of those in need received antiretroviral therapy (ART).1 The cost of combination ART has dropped in recent years as a result of generic medicines and differential pricing based on country need and ability to pay.2 The cost of diagnostic services to determine eligibility for treatment and to monitor treatment response has kept ART inaccessible to many, however. The US Department of Health and Human Services (DHHS) and the World Health Organization (WHO) recommend initiating ART therapy based on consideration of a patient's CD4 T-cell count when available.3,4 A CD4 cell count requires expensive laboratory equipment and trained technicians, which are absent in many areas of high HIV prevalence. In April 2002, the WHO published alternative recommendations for initiation of ART in settings where CD4 testing is unavailable based on the WHO staging system for HIV infection (Table 1) and the total lymphocyte count (TLC). In December 2003, the WHO broadened the recommendations for initiation of ART when CD4 testing is unavailable to include WHO stage III or IV or WHO stage II in combination with a TLC ≤1200 cells/mm3.4
Recent literature provides support for accessible inexpensive laboratory measures as surrogates for CD4 count in determining the need for ART. Several studies have shown a good correlation between the TLC and CD4 count5-9 and have documented similar sensitivity of these measures in predicting HIV mortality.10,11 The utility of the TLC for predicting a low CD4 count has been found to improve when considered in combination with the hemoglobin level.5,12 Anemia, which is frequent in persons with HIV infection,13,14 predicts clinical prognosis independent of CD4 count and viral load.14,15 Other potential inexpensive markers of late-stage disease have been reported but have not been consistent. We examined inexpensive indices of immunosuppression for their sensitivity and specificity in detecting a CD4 count <200 cells/mm3 and a CD4 count <350 cells/mm3 to determine whether the current recommendations for initiation of ART in resource-limited settings could be improved.
A study of the heterosexual transmission of HIV was initiated in 1992 in Chiang Mai, Thailand,16 where HIV-1 circulating recombinant form (CRF) 01_AE (subtype E) was the predominant subtype. By March 1997, 590 seropositive male blood donors and their female partners, 276 (46.8%) of whom were HIV-positive, had enrolled in the study. At enrollment, eligible subjects had a blood specimen obtained together with an interview and a physical examination. Serum samples were tested by HIV-1 enzyme-linked immunosorbent assay (ELISA) and, if positive, by Western blot analysis. Numerous other blood tests were performed for a variety of markers, including immune markers and viral subtype. Because of the low prevalence of subtype B infections in this cohort, our current analysis is limited to participants with subtype E virus and includes 839 of 866 HIV-positive subjects. The current analysis was limited to participants for whom a TLC, complete blood cell count (CBC), and CD4 count measured by flow cytometry were available and for whom all enrollment data were collected within a 6-week period.
For each participant, the WHO clinical stage at enrollment was assigned by one of the authors (D. J. Jamieson) based on the clinical data, including the physical examination, and the reported history of specific symptoms (hospitalization reasons, oral thrush, fever lasting at least 7 days, diarrhea lasting at least 2 weeks, or unexplained weight loss of at least 5 kg). The combination of WHO stage and TLC determined whether or not each subject met the WHO guidelines for ART in resource-limited settings. The WHO staging system for HIV in adults and adolescents is shown in Table 1. We assessed the April 2002 WHO guidelines (WHO stage IV or a TLC <1200 cells/mm3 in combination with WHO stage II or III) and the December 2003 guidelines (WHO stage III or IV or the combination of WHO stage 2 and a TLC ≤1200 cells/mm3). The TLC was assessed for independent predictive value at varying cell count cutoffs. Anemia was defined as a hemoglobin level <12 g/dL for men and <11 g/dL for women. Body mass index (BMI) was calculated from height and weight obtained during the physical examination; however, BMI was missing in 17.2% of the men and 13.9% of the women. Subjects missing physical examination data were not classified into WHO stage or according to the 2002 or 2003 WHO guidelines for initiation of ART.
The correlation between CD4 cell count and TLC among HIV-positive men and women was assessed using scatterplots and Spearman correlation coefficients. Stratified by gender, potential predictors were assessed for their sensitivity and specificity for a CD4 count <200 cells/mm3 and a CD4 count <350 cells/mm.3 The association was statistically tested using the Mantel-Haenszel χ2 test. Receiver operating characteristic (ROC) curves were used to visualize sensitivity and specificity of incremental cutoffs of the TLC and hemoglobin level by gender. Computer-generated combinations of TLC cutoffs, anemia, and WHO stage of HIV infection were assessed for improved sensitivity for a CD4 count <200 cells/mm3 and a CD4 count <350 cells/mm3 compared with individual predictors. Multivariate logistic regression was used to explore the addition of potential predictors to the current WHO recommendations for initiation of ART in resource-limited settings where CD4 testing is unavailable. Potential novel resource-limited algorithms were compared with the current recommendations for their sensitivity and specificity for predicting a low CD4 count.
The median age of the HIV-positive men and women was 29 (range: 18-48) and 26 (range: 15-52) years, respectively. None of the women and only 2.1% (n = 11) of the men had a history of cocaine or opiate use within the 3 years before enrollment. At enrollment, all subjects were antiretroviral naive. Men, who were infected before their female partner as defined by enrollment criteria, had signs, symptoms, and laboratory indicators consistent with more advanced HIV disease at enrollment (Table 2). A higher proportion of the men than the women had a CD4 count <200 cells/mm3 (30.2% vs. 11.1%, respectively).
An analysis of the continuous TLC and CD4 count found a high correlation (men: Spearman R = 0.68, P < 0.001; women: Spearman R = 0.72, P < 0.001; Fig. 1). ROC curves for TLC cutoffs indicate that a TLC <1500 cells/mm3 approximately doubled the sensitivity for a CD4 count <200 cells/mm3 as compared with a TLC <1200 cells/mm3, with minimal loss of specificity for both genders (Fig. 2A). Similar ROC curves for a CD4 count <350 cells/mm3 (see Fig. 2B) show increased sensitivity of a higher TLC cutoff (ie, TLC <1700 cells/mm3 or TLC <1800 cells/mm3) for a CD4 count <350 cells/mm3 while maintaining similar specificity as a TLC <1500 cells/mm3 for a CD4 count <200 cells/mm3.
ROC curves of hemoglobin level cutoffs for predicting a CD4 count <200 cells/mm3 substantiate the increased severity of anemia with progressing disease (see Fig. 2C). A comparison of the male and female ROC curves suggests that the greatest balance of sensitivity and specificity was at hemoglobin cutoffs consistent with moderate to severe anemia for men and consistent with severe anemia for women. The ROC curve of hemoglobin cutoff for predicting a CD4 count <350 cells/mm3 (data not shown) had a similar pattern and specificity to that for a CD4 count <200 cells/mm3. The addition of TLC cutoffs to anemia improved the predictability of anemia in women for a CD4 count <200 cells/mm3 (see Fig. 2D). Among women, anemia and a TLC <2000 cells/mm3 had a sensitivity of 16.7% and specificity of 98.1% for a CD4 count <200 cells/mm3.
The original WHO recommendations for initiation of ART when CD4 testing is unavailable had high specificity for a CD4 count <200 cells/mm3 (99.7% for men, 99.5% for women) but rather low sensitivity (17.8% for men, 9.1% for women; see Table 2). The December 2003 revision more than doubled the sensitivity for a CD4 count <200 cells/mm3 compared with that of the original WHO recommendations (34.1% for men, 31.8% for women), albeit with a loss of specificity (90.2% for men, 86.0% for women). A TLC <1500 cells/mm3 had a similar sensitivity (43.0% for men, 33.3% for women) to the December 2003 recommendations but with higher specificity (97.2% for men, 95.8% for women). Anemia had greater than 96% specificity for a CD4 count <200 cells/mm3 and a CD4 count <350 cells/mm3 among men and 85% specificity for a CD4 count <200 cells/mm3 and a CD4 count <350 cells/mm3 among women.11 In an exploratory stepwise logistic regression analysis with the inclusion of the December 2003 WHO recommendations and a TLC <1200 cells/mm3, a TLC <1500 cells/mm3, a TLC <1800 cells/mm3, anemia, and BMI <18.5 kg/mm2 in the model, TLC <1500 cells/mm3 and anemia were selected as additional predictors for a CD4 count <200 cells/mm3 for men and women (see Table 2). For a CD4 count <350 cells/mm3, a TLC <1800 cells/mm3 was selected alone as an additional predictor to the current guidelines.
An algorithm that uses only the TLC and hemoglobin level (TLC <1500 cells/mm3 or TLC <2000 cells/mm3 with anemia) had higher sensitivity and specificity than the current WHO guidelines for a CD4 count <200 cells/mm3 for men and women (Table 3). The addition of a TLC <1500 cells/mm3 and a TLC <2000 cells/mm3 with anemia to the December 2003 revised WHO recommendations substantially increased the sensitivity for detecting a CD4 count <200 cells/mm3 (60.7% for men, 63.6% for women) compared with that of the current guidelines alone (34.1% for men, 31.8% for women), with a minimal decrease in specificity (−4.7% for men, −6.0% for women; see Table 3). Including a TLC <2000 cells/mm3 in combination with WHO stage II of HIV infection to the algorithm slightly increased the sensitivity (63.0% for men, 68.2% for women) with little impact on specificity (−2.0% for men, −0% for women).
For a CD4 count <350 cells/mm3, an algorithm with a higher TLC cutoff at 1700 cells/mm3 along with a combination of a TLC <2000 cells/mm3 and anemia provided improved sensitivity and specificity compared with those of the current guidelines for a CD4 count <350 cells/mm3 among men and women. When a TLC <1700 cells/mm3 and the combination of a TLC <2000 cells/mm3 with anemia are added to the algorithm, including the current WHO guidelines, the sensitivity for a CD4 count <350 cells/mm3 (46.0% for men, 45.1% for women) was double that of the current guidelines (24.0% for men, 23.2% for women) with little decrease in specificity (−2.7% for men, −2.4% for women). The addition of the combination of a TLC <2000 cells/mm3 and WHO stage II of HIV infection had little effect on the December 2003 WHO algorithm's sensitivity and specificity for a CD4 count <350 cells/mm3.12 All proposed algorithms maintained a median CD4 count among misclassified subjects that was less than the median CD4 count among subjects misclassified by the current guidelines.
If CD4 lymphocyte counts were available and an HIV-infected patient's CD4 lymphocyte count was less than 200 cells/mm3, it would be universally agreed that ART should be initiated regardless of the fact that CD4 counts have biologic variability by sex, age, and ethnicity as well as some variability in measurement in the laboratory. Although the CD4 count has become the “gold standard,” in resource-limited settings where the CD4 count is not routinely available, other clinical and laboratory factors can be predictive of severe immune compromise. Routine hematologic testing is inexpensive and widely accessible and can provide measurements of the TLC and hemoglobin level. The TLC was highly correlated with the CD4 count in this Thai cohort, consistent with studies of HIV-infected patients in South Africa,6 the United Kingdom,7 and the United States.5,8,9 The standard TLC cutoff of <1200 cells/mm3 was highly specific for a CD4 count <200 cells/mm3 in our population, but the TLC cutoff of <1500 cells/mm3 had a better balance of sensitivity and specificity, allowing almost twice as many subjects with a CD4 count <200 cells/mm3 to be identified. The higher TLC cutoff retained an acceptable level of specificity.
Among the men in this study, anemia was highly predictive of a CD4 count <200 cells/mm3. Among the women, who had a higher prevalence of anemia, we found that in addition to being anemic, having a TLC <2000 cells/mm3 greatly increased the specificity of anemia for predicting a CD4 count <200 cells/mm3. Spacek et al5 suggested the combination of anemia with a TLC from 1200 to 2000 cells/mm3 in addition to a TLC <1200 cells/mm3 for increased sensitivity and specificity for a CD4 count <200 cells/mm3 in a US population. Several studies have shown a consistently higher prevalence of anemia in patients with AIDS.13,14 The largest analysis of anemia in HIV infection to date found the prevalence of anemia to be 46% to 59% higher among HIV-infected patients with AIDS than in those without AIDS.13
Combining anemia and a TLC algorithm with the current WHO recommendations for initiation of ART substantially increases the sensitivity for a CD4 count <200 cells/mm3 and a CD4 count <350 cells/mm3 compared with either algorithm alone. In our population, the addition of a moderate TLC cutoff (TLC <1500 cells/mm3 for a CD4 count <200 cells/mm3 and TLC <1700 cells/mm3 for a CD4 count <350 cells/mm3) and anemia with a TLC <2000 cells/mm3 doubled the sensitivity compared with that of the current WHO recommendations.
This study has many strengths, including the size and characteristics of the cohort, which comprised antiretroviral-naive men and women with heterosexually acquired HIV, along with the abundance of clinical and laboratory data obtained at enrollment. Nevertheless, there are some limitations. Enrollment physical examination data were missing for 89 (16.2%) men and 30 (11.9%) women. Because the physical examination data were used in conjunction with the questionnaire data to determine WHO disease stage, a small number of subjects could not be classified according to WHO stages of HIV infection. When the subjects with missing physical examination data were included, there was no significant difference in the sensitivity of revised WHO recommendations for onset of ART (men: data included for 15.7% and data excluded for 17.5%; women: data included for 7.1% and data excluded for 8.0%). This is likely a result of the low prevalence of late-stage symptoms in our population. We believe that the missing data should not affect the conclusions drawn from this analysis. Another potential limitation is the generalizability of the results from this subtype E-infected heterosexual Thai population. Reports from other nations, including the United States and South Africa, support the utility of the TLC and anemia for increasing the sensitivity for identifying individuals with a CD4 count <200 cells/mm3.5-12,14,15
The December 2003 revision of the WHO guidelines for initiation of ART doubled the sensitivity to detect persons with a CD4 count <200 cells/mm3 compared with that of the previous guidelines. The sensitivity could potentially be doubled further with minimal loss to specificity if the current recommendations included a TLC <1500 cells/mm3 and the combination of a TLC <2000 cells/mm3 with anemia or WHO stage II HIV infection. This algorithm maintained high specificity and applicability among men and women who were at different stages of HIV disease.
The authors thank Sonya Bowens and Antika Wongthanee for their data management contributions to the Thai heterosexual couples study.
1. Joint United Nations Programme on HIV/AIDS (UNAIDS). World Health Organization (WHO). AIDS Epidemic Update
. UNAIDS/WHO, Geneva, Switzerland, 2003.
2. Joint United Nations Programme on HIV/AIDS (UNAIDS). Report on the global HIV/AIDS epidemic 2002-Barcelona report. UNAIDS, Geneva, Switzerland, 2002.
5. Spacek LA, Griswold M, Quinn TC, et al. Total lymphocyte count
and hemoglobin combined in an algorithm to initiate the use of highly active antiretroviral therapy in resource-limited settings. AIDS
6. Van der Ryst E, Kotze M, Joubert G, et al. Correlation among total lymphocyte count
, absolute CD4+ count, and CD4 percentage in a group of HIV-1
-infected South African patients. J. AIDS Hum Retroviruses
7. Beck EJ, Kupek CJ, Gompels MM, et al. Correlation between total and CD4 lymphocyte counts in HIV infection: not making the good an enemy of the not so perfect. Int J STD AIDS
8. Blatt SP, Lucey CR, Butzin CA, et al. Total lymphocyte count
as a predictor of absolute CD4+ count and CD4+ percentage in HIV-infected persons. JAMA
9. Fournier AM, Sosenko LM. The relationship of total lymphocyte count
to CDC lymphocyte counts in patients infected with human immunodeficiency virus. Am J Med Sci
10. Bedell R, Heath KV, Hogg RS, et al. Total lymphocyte count
as a possible surrogate of CD4 cell count to prioritize eligibility for antiretroviral therapy among HIV-infected individuals in resource-limited settings. Antivir Ther
11. Post FA, Wood R, Maartens G. CD4 and total lymphocyte counts as predictors of HIV disease progression. Q J Med
12. Gange SJ, Lau B, Phair JP, et al. Use of total lymphocyte count
and hemoglobin for staging HIV disease and possibly initiating antiretroviral therapy [abstract TuPeC4706]. Presented at: XIVth International AIDS Conference; 2002; Barcelona.
13. Sullivan PS, Hanson DL, Chu SY, et al. Adult/Adolescent Spectrum of Disease Group. Epidemiology of anemia
in human immunodeficiency virus (HIV)-infected persons: results from the multistate Adult and Adolescent Spectrum of HIV Disease Surveillance Project. Blood
14. Belperio PS, Rhew DC. Prevalence and outcomes of anemia
in individuals with human immunodeficiency virus: a systematic review of the literature. Am J Med
15. Mocroft A, Kirk O, Barton SE, et al. Anaemia is an independent predictive marker for clinical prognosis in HIV-infected patients from across Europe. AIDS
16. Nagachinta T, Duerr A, Suriyanon V, et al. Risk factors for HIV-1
transmission from HIV-seropositive male blood donors to their regular female partners in northern Thailand