Portal de Periódicos da CAPES

Absolute or total lymphocyte count as a marker for the CD4 T lymphocyte criterion for initiating antiretroviral therapy

2003; Lippincott Williams & Wilkins; Volume: 17; Issue: 6 Linguagem: Inglês

10.1097/00002030-200304110-00019

1473-5571

Mark Z. Jacobson, Lea Liu, H Khayam-Bashi, Steven G. Deeks, Frederick Hecht, James O. Kahn,

HIV-related health complications and treatments

We analysed simultaneous CD4, absolute (ALC) and total (TLC) lymphocyte counts of 2057 patients. ALC of 1750 cells/μl or less optimized sensitivity and specificity for predicting CD4 cell counts less than 350 cells/μl. ALC sensitivity/specificity tended to be higher for blacks, patients under 50 years old, and patients with higher plasma HIV-RNA values. TLC values slightly higher than ALC provided virtually identical results. ALC or TLC may be helpful in deciding when to initiate antiretroviral therapy in resource-poor settings. As highly active antiretroviral therapy (HAART) is now becoming available to large populations of HIV-infected patients in resource-poor countries, resource-appropriate markers need to be identified for clinicians to use in deciding when to initiate HAART. Previous studies suggested the absolute lymphocyte count (ALC) or total lymphocyte count (TLC, i.e. ALC plus all large lymphocytes such as lymphoblasts or reactive lymphocytes) might be useful in identifying patients who would benefit from initiating prophylaxis for AIDS-related opportunistic infections [1–3]. We therefore conducted a study to determine whether ALC or TLC has the potential clinical utility in identifying patients who would benefit from HAART, analysing sensitivity and specificity over a range of ALC and TLC values for a CD4 cell count of either less than 350 or less than 200 cells/μl (values chosen because of the consensus that HAART should be deferred until a patient's CD4 cell count drops below 350 cells/μl, but should be initiated before the CD4 cell count drops below 200 cells/μl) [4–10]. We examined how factors not previously examined (sex, race, age, HIV viral load, antiretroviral treatment) might affect the performance of ALC or TLC as a surrogate for absolute CD4 T lymphocyte counts. The patients studied included all individuals receiving care in the Positive Health Program HIV clinic at San Francisco General Hospital. Data were obtained from HIVWatch, the electronic medical record system used in the clinic. One simultaneous absolute CD4 T lymphocyte count, ALC and TLC was randomly chosen for analysis from each patient. Complete blood counts with differential were performed by a Bayer Advia 120 Hematology Analyzer (Bayer Diagnostics, Tarrytown, NY, USA), and absolute CD4 T lymphocyte counts were measured by a three-color, flow cytometry, cell counter (BD Immunodiagnostics, San Jose, CA, USA). A series of receiver operator characteristic (ROC) curves, which graph sensitivity as a function of (1 − specificity), were created for each group and subgroup over the entire range of ALC and TLC values for an absolute CD4 cell count of less than 350 cells/μl or less than 200 cells/μl. Sensitivity, specificity, and positive and negative predictive values were calculated for a range of ALC or TLC values. The area under the ROC curve (AUC) for each subgroup examined was calculated to assess the accuracy of the classification. A total of 2356 patients had at least one absolute CD4 T lymphocyte count measured between 1 January and 31 December 2001, of whom 2057 had one or more simultaneous ALC and TLC and complete demographic data (85.5% male, 14.5% female; 50.9% white, 30.7% black, and 13.5% Hispanic; 81.8% ≤ 50 years and 18.3% > 50 years of age; 31, 19 and 50% had absolute CD4 T cells counts < 200, 200–350 and > 350 cells/μl, respectively). A subset of 1356 patients also had plasma HIV-RNA levels measured on the same date (31.2, 29.7 and 39.1% had plasma HIV-RNA levels < 50, 50–10 000 and > 10 000 copies/ml), and 978 patients could also be categorized as being on (89%) or off (11%) antiretroviral treatment at the time of a CD4/ALC/TLC cell count. The correlation of ALC and a CD4 cell count, and of TLC and a CD4 cell count were highly significant (correlation coefficients of 0.69 and 0.68, respectively, P < 0.001 for each). An ALC value of 1750 cells/μl or less had maximal combined sensitivity, 74% [95% confidence interval (CI) 72–76%], and specificity, 73% (95% CI 72–75%), for a CD4 cell count of less than 350 cells/μl (Table 1). An ALC of 1500 cells/μl or less had maximal combined sensitivity, 79% (95% CI 77–81%), and specificity, 75% (95% CI 73–77%), for a CD4 cell count of less than 200 cells/μl. TLC cutoff values that were 150 cells/μl higher than the ALC value had virtually identical combined sensitivity and specificity. Notably, an ALC of less than 1000 and a TLC of less than 1150 cells/μl each had a 98% positive predictive value for a CD4 cell count of less than 350 cellsμl; an ALC of less than 750 cells/μl and a TLC of less than 900 cells/μl had a positive predictive value of 96 and 94%, respectively, for a CD4 cell count of less than 200 cells/μl.Table 1: Combined sensitivity, specificity, positive predictive value, negative predictive value and likelihood ratio of absolute lymphocyte count for absolute CD4 T lymphocyte count of less than 350 cells/μl.In subgroup analyses, ROC curves did not differ significantly by sex. There was a trend towards higher overall combined sensitivity and specificity of ALC and TLC for black than for white individuals, for patients 50 years old or less than for those over 50 years old, and for patients with detectable HIV-RNA loads. For example, considering ALC as a marker for CD4 cell counts of less than 350 cells/μl, the ROC AUC was 0.852 (95% CI 0.823–0.881) among black versus 0.799 (0.733–0.826) among white individuals. Among patients for whom antiretroviral treatment status could be accurately characterized at the time of sampling, there were 513 patients with a CD4 cell count of less than 350 cells/μl (60 not on antiretroviral treatment and 453 on treatment) and 467 patients with a CD4 cell count of 350 cells/μl or greater (48 not on treatment and 419 on treatment). Considering ALC as a marker for CD4 cell counts less than 350 cells/μl, the ROC AUC was 0.822 (95% CI 0.745–0.899) for those not on antiretroviral treatment versus 0.816 (95% CI 0.789–0.844) for those on treatment. TLC analysis showed similarly overlapping ROC AUC values for those not on versus those on treatment. A critical issue is the proportion of patients with a CD4 cell count of less than 200 cells/μl who would be misclassified by ALC as having a CD4 cell count greater than 350 cells/μl, and who would thus mistakenly have antiretroviral treatment deferred. Using an ALC of 1750 cells/μl or less as the threshold to initiate HAART, 82 of the 605 patients (14%) with a CD4 cell count of less than 200 cells/μl would have been misclassified, by having a simultaneous ALC greater than 1750 cells/μl, as not being in need of antiretroviral treatment. Of note is the fact that only 22 of these 605 (3.6%) patients had a CD4 cell count of less than 100 cells/μl, and although at very high risk of opportunistic infection, would have been misclassified by an ALC greater than 1750 cells/μl as not needing antiretroviral treatment. Lower ALC cutoff values increased the percentage of patients with a CD4 cell count of less than 200 cells/μl who would be misclassified as not requiring treatment, and decreased the proportion of patients with CD4 cell counts of 350 cells/μl or greater who would have been unnecessarily treated. In summary, our findings suggest that monitoring either ALC or TLC could have clinical utility as a basis for determining when HIV-infected patients in resource-poor settings should initiate HAART. In particular, ALC and TLC cutoff values could be defined that had such a high PPV for a CD4 cell count of less than 350 cells/μl that CD4 cell enumeration might not be needed to confirm the need for antiretroviral treatment. ALC oC or TLC cutoff values with similar predictive values for lower CD4 cell count criteria could also be identified if le less than 350 cells/μl is not the optimal value in a given setting. Future studies should also examine whether the trajectory of ALC or TLC in response to HAART has potential clinical utility for determining when patients in resource-poor countries are failing antiretroviral treatment and should switch to a salvage HAART regimen.