ABSTRACT
Aim:
To find a sensitive and low-cost surrogate marker for CD4 count for initiating highly active antiretroviral therapy (HAART) [CD4 < 200 /mm3], in the form of total lymphocyte count (TLC) < 1200 /mm3 combined with haemoglobin (Hb) with multiple Hb cut-offs.
Method:
Two hundred and three consecutive treatment-naïve adult HIV positive outpatients attending the virology clinic in World Health Organization (WHO) clinical stage 1, 2 or 3 were enrolled in the study. Their complete blood counts and CD4 counts were done. Descriptive statistics was done by two methods correlating TLC alone with CD4 and the other using combined marker of TLC and Hb with CD4 count.
Result:
Total lymphocyte count alone did not correlate well with CD4 counts (r = 0.13; p = 0.065). Sensitivity of TLC < 1200 /mm3 to predict CD4 < 200 /mm3 was low (23.27%) and the sensitivity of the combined marker (TLC + Hb) increased with higher Hb cut-offs.
Conclusion:
Adding Hb to TLC markedly improved the sensitivity of the marker to predict CD4 count < 200/mm3. We also recommend a trade-off Hb cut-off of 10.5 g/dL for optimum sensitivity and specificity in this population subset.
Keywords: Haemoglobin, HIV, surrogate marker, total lymphocyte count
RESUMEN
Objetivo:
Encontrar un marcador sustituto sensible y de bajo costo para el conteo de CD4 para iniciar la terapia antirretroviral altamente activa (HAART) [CD4 < 200 /mm3], en forma de conteo total de linfocitos (CTL) < 1200 /mm3 combinado con la hemoglobina (Hb) con múltiples valores de corte de Hb.
Método:
Doscientos y tres pacientes adultos ambulatorios consecutivos, con SIDA pero sin tratamiento previo, que asistían a la clínica de virología en las etapas clínicas 1, 2, ó 3 de la Organización Mundial de la salud (OMS) 1, 2 ó 3, fueron captados para este estudio. Se les realizaron conteos sanguíneos y conteos de CD4 completos. Se hizo una estadística descriptiva mediante dos métodos: uno que correlacionaba solo el CTL con el CD4, y el otro que combinaba el marcador de CTL con el conteo de CD4.
Resultado:
El conteo total de linfocitos solo no tuvo una buena correlación con los conteos de CD4 (r = 0.13; p = 0. 065). La sensibilidad del CTL < 1200 /mm3 para predecir CD4 < 200/mm3 fue baja (23.27%) y la sensibilidad del marcador combinado (CTL + Hb) aumentó en la medida que los valores de corte de Hb fueron mayores.
Conclusión:
La adición de Hb al CTL mejoró notablemente la sensibilidad del marcador para predecir el conteo de CD4 < 200/mm3. También recomendamos un valor de corte de Hb negociable de 10.5 g/dL, para una óptima sensibilidad y especificidad en este subgrupo de población.
INTRODUCTION
Human immunodeficiency virus (HIV) chiefly affects and multiplies within CD4 cells, mainly CD4+ TH cells. Hence, CD4 cell counts are said to corroborate with the disease progression (1, 2). CD4+ TH lymphocytes activate all other cell lines of immunity by means of various releasing factors. Therefore, a fall in CD4 cell count affects other cell lines in the immune system. This lapse in the immunity is the cause of opportunistic infections which form the basis of the World Health Organization (WHO) clinical classification. The patients in WHO clinical stage 4 qualify for initiation of highly active antiretroviral therapy (HAART), irrespective of blood counts. We have considered only patients in WHO clinical stages 1, 2 or 3, who may be deprived of HAART in a resource-limited setting. The use of HAART is limited by the cost of the therapy and also the risk of development of drug resistance. Hence, HAART is initiated when CD4 count < 200/mm3 as per Centers for Disease Control and Prevention (CDC) guidelines or when total lymphocyte count (TLC) < 1200/mm3 as per the WHO. Many studies have found a strong correlation between TLC and CD4 counts (3–6), while a few studies had contrary findings (7–9). Eventually, haemoglobin (Hb) concentration was proved to be an independent prognostic marker of AIDS (3, 10). The aim of this study was to test the cheaper surrogate markers of CD4 counts as TLC and Hb independently as well as to better the conventional TLC surrogate marker by designing a novel algorithm in a subset of the Indian population. Our algorithm differs from other studies in using several Hb cut-offs in this Indian population subset.
SUBJECTS AND METHODS
After approval from the Institutional Ethics Committee of Seth GS Medical College and KEM Hospital, 203 consecutive treatment-naïve seropositive adult patients attending the virology clinic at a tertiary healthcare centre in Mumbai were included. At presentation, patients were staged clinically as per the revised WHO clinical staging recommendations. This was done on the basis of general examination and previous reports of patients from their respective referral centres. Patients already on HAART and in WHO clinical stage 4 were excluded since these patients were recommended to be put on HAART irrespective of their CD4 counts (7). The remaining patients were investigated for their complete blood count including the total lymphocyte count and haemoglobin concentration and CD4 cell counts by flow cytometry. Two algorithms were used to predict CD4 count < 200/mm3 (Figure). In the first algorithm, only TLC was used as a marker, and a correlation between CD4 counts and TLC was performed. Then, we used a single TLC cut-off of 1200/mm3 to predict an outcome of CD4 count < 200/mm3. Hence, TLC < 1200/mm3 was defined as a positive test outcome and TLC ≥ 1200/mm3 was labelled a negative test outcome. These data were then analysed to obtain descriptive statistics. Next, we performed a correlation between CD4 count and Hb. In the second algorithm, we intended to better the sensitivity of the earlier algorithm using Hb as a parameter. To facilitate the same, the positive outcome of this new algorithm was now determined as a summation of the group of positive test outcomes with TLC < 1200/mm3 in the first algorithm and also another group of patients with TLC ≥ 1200/mm3 but with Hb less than a cut-off value. This Hb cut-off was serially graded as 9, 9.5, 10, 10.5, 11, 11.5, 12, 12.5 g/dL.
Figure. The surrogate marker algorithms.
As an example, if an Hb cut-off of 9 g/dL was selected as per the second algorithm, the new set of positive outcomes to determine CD4 count of 200/mm3 would be a total of the patients with TLC < 1200/mm3 and patients with TLC ≥ 1200 but Hb concentration < 9 g/dL. A descriptive analysis was done with this new algorithm with various serial Hb cut-off values.
The data from a total of 203 treatment-naïve seropositive patients were analysed using SPSS version 10 with level of significance being 5% and power of 80%. Pearson's correlation coefficient was used to find the correlation between TLC and CD4 and then between Hb and CD4 counts. Then a 2 × 2 contingency table was used to determine the descriptive analysis like the sensitivity, specificity, positive predictive value, and to compare the two algorithms.
RESULTS
The mean age of the 203 participants was 33.35 (± 8.15) years. Sixty-six (33%) were females and the remaining 134 (67%) were males. The mean CD4 count was 224.35/mm3 (± 174.3) and that of TLC was 1839.22/mm3 (± 741.16). On comparing the CD4 and TLC alone, the Pearson's coefficient of correlation (r) was found to be 0.130 and coefficient of variance was 0.017, p-value being 0.0671 (two-tailed). The mean haemoglobin value was 9.78 (± 1.62) g/dL. The Pearson's coefficient of correlation (r) was 0.273 and coefficient of variance was 0.075 with p-value being 0.0001 (two-tailed).
The data were analysed using the 2 × 2 contingency table. The results of sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of TLC alone, as well as TLC combined with Hb concentration to predict a CD4 count < 200/mm3 using both the algorithms are shown in the Table. The sensitivity of the first algorithm using TLC alone was 23.27% and specificity was 86.90% and that of the second algorithm using Hb cut-off of 10.5 g/dL was 80.17% and specificity was 40.47%. The PPV remained low in both cases (71.05% and 65.03%, respectively).
Table. Performance of surrogate marker with serially graded haemoglobin cut-off level.
| n = 203 CD4 count < 200/mm3 | Sensitivity (%) | Specificity (%) | PPV (%) | NPV (%) |
|---|---|---|---|---|
| First method: WHO criterion TLC < 1200/mm3 | 23.27 | 86.90 | 71.05 | 45.06 |
| Second method: combined with Hb | ||||
| a) TLC < 1200/mm3 (as in first method) and (TLC ≥ 1200/mm3 but Hb < 9 g/dL) | 49.13 | 73.80 | 72.15 | 51.23 |
| b) TLC < 1200/mm3 (as in first method) and (TLC ≥ 1200/mm3 but Hb < 9.5 g/dL) | 54.31 | 61.90 | 66.31 | 49.52 |
| c) TLC < 1200/mm3 (as in first method) and (TLC ≥ 1200/mm3 but Hb < 10 g/dL) | 64.28 | 57.14 | 66.67 | 52.17 |
| d) TLC < 1200/mm3 (as in first method) and (TLC ≥ 1200/mm3 but Hb < 10.5 g/dL) | 80.17 | 40.47 | 65.03 | 59.64 |
| e) TLC < 1200/mm3 (as in first method) and (TLC ≥ 1200/mm3 but Hb < 11 g/dL) | 83.62 | 28.57 | 61.78 | 55.81 |
| f) TLC < 1200/mm3 (as in first method) and (TLC ≥ 1200/mm3 but Hb < 11.5 g/dL) | 89.65 | 17.85 | 60.11 | 55.56 |
| g) TLC < 1200/mm3 (as in first method) and (TLC ≥ 1200/mm3 but Hb < 12 g/dL) | 92.24 | 11.90 | 59.11 | 52.63 |
| h) TLC < 1200/mm3 (as in first method) and (TLC ≥ 1200/mm3 but Hb < 12.5 g/dL) | 99.13 | 8.34 | 59.89 | 87.5 |
PPV = positive predictive value, NPV = negative predictive value, WHO = World Health Organization, Hb = haemoglobin, TLC = total lymphocyte count
DISCUSSION
Highly active antiretroviral therapy is to be initiated in an HIV-positive patient with a CD4 count < 200/mm3. This cutoff is selected since HAART is limited by the cost of the therapy and also the risk of development of drug resistance if initiated prematurely, and risks of disease complications if HAART is started very late. These financial issues are compounded in resource-limited settings where CD4 cell counts are cost-prohibitive. Hence, a cheaper surrogate for CD4 cell count would benefit the resource-limited settings. With conflicting views in different studies regarding the use of only TLC as a CD4 count surrogate marker, we came across another reliable, cheap and easily available marker such as Hb concentration (3–10). We designed an algorithm combining both these surrogate markers to determine a CD4 count < 200/mm3.
In this study, the sensitivity of the first algorithm using TLC < 1200/mm3 alone to determine CD4 count < 200/mm3 was 23.27% and specificity was 86.90%. This would deprive many (76.73%) deserving patients from receiving HAART. There is a need to improve the sensitivity of this marker. Most studies used various graded TLC cut-offs, though we did not use the same in this study (3, 7, 11).
In a study by Spacek et al, a similar TLC and Hb combined surrogate marker was used to determine CD4 count < 200/mm3 (7). This study used two different TLC cut-offs (1200 and 2000/mm3) as against a single TLC (1200/mm3) cut-off used in our study. Spacek et al also included patients who were on HAART, measured their serial cell counts and used four different methods with frequent use of flow cytometry, which we thought was not practical to implement in a resource-limited setting where frequent use of flow cytometry is cost-prohibitive. Hence, we chose to include only treatment-naïve patients, with flow cytometry being used only once for CD4 cell counts and we designed a very simple, easy to understand algorithm.
The other issue was to select a suitable trade-off for optimum sensitivity and specificity. A test with poor sensitivity will deprive many needy patients of HAART and one with a very high sensitivity will be cost prohibitive and also raise the issue of drug resistance. In the study by Spacek et al, a single Hb cut-off of 12 g/dL was used. We instead selected serial graded Hb cut-offs in the second algorithm from which we recommend using a cut-off of 10.5 g/dL in this population subset to optimize the sensitivity (80.17%), specificity and PPV. This finding is in concert with the study by Spacek et al which states that the sensitivity of the marker can be improved by adding Hb as a parameter and that each setting must define its own algorithm to cater to a different subset of population. As a corollary, even in an economically sound setting, CD4 count could be done selectively for patients with Hb < 10.5 g/dL (or a cut-off suitable for that subset of population) without performing the TLC, thereby reducing investigational costs.
We conducted this study in a specific subset of the population. There are many confounding factors such as malnutrition, concurrent subclinical infections, gender and physiological variations which affect both TLC and Hb (12). The incidence of anaemia depends on nutritional status, ethnicity, gender and pregnancy. Hence, there could be a variation in these surrogate parameters among the developing and developed countries. Though we cannot generalize the findings of this study since it is limited to this population subset, we do recommend that the sensitivity of the conventional surrogate marker can be improved with newer cost-effective algorithms. Similar algorithms should be designed for a larger sample size and different population subsets. More efforts are needed to scale-up access to CD4 count. Also, we do understand that CD4 count itself is a surrogate marker for the viral load (13).
CONCLUSION
Total lymphocyte count alone does not seem to correlate with the CD4 count. Haemoglobin does seem to be an important predictor of disease activity. Adding Hb to TLC does improve the sensitivity of the marker to predict the CD4 count < 200/mm3, ensuring initiation of HAART and thereby reducing investigational cost. A trade-off of a Hb cut-off of 10.5 g/dL is suggested to be used to optimize the use of this surrogate marker in this population subset. The sensitivity of using TLC alone was 23.27% and specificity was 86.90%, but while also using Hb cut-off of 10.5 g/dL, sensitivity was 80.17% and specificity was 40.47%. The PPV remained low in both cases (71.05% and 65.03%, respectively). There is a glaring need to scale-up access to CD4 count availability to accurately initiate and monitor HAART.
ACKNOWLEDGEMENT
The authors wish to thank Dr Nithya J Gogtay, Addl Professor, Department of Clinical Pharmacology, Seth GSMC and KEM Hospital, Parel, for her continuous support and consistent encouragement.
Footnotes
The results of this study were presented in 2005 to the Indian Council of Medical Research, New Delhi, India, as a short-term research.
REFERENCES
- 1.Egger M, May M, Chêne G, Phillips AN, Ledergerber B, Dabis F, et al. Prognosis of HIV-1-infected patients starting highly active antiretroviral therapy: a collaborative analysis of prospective studies. Lancet. 2002;360:119–129. doi: 10.1016/s0140-6736(02)09411-4. [DOI] [PubMed] [Google Scholar]
- 2.Hammer SM, Squires KE, Hughes MD, Grimes JM, Demeter LM, Currier JS, et al. A controlled trial of two nucleoside analogues plus indinavir in persons with immunodeficiency virus infection and CD4 cell counts of 200 per cubic millimeter or less. AIDS Clinical Trials Group 320 Study Team. N Engl J Med. 1997;337:725–733. doi: 10.1056/NEJM199709113371101. [DOI] [PubMed] [Google Scholar]
- 3.Mocroft A, Kirk O, Barton SE, Dietrich M, Proenca R, Colebunders R, et al. Anaemia is an independent predictive marker for clinical prognosis in HIV infected patients from across Europe. Euro SIDA study group. AIDS. 1999;13:943–950. doi: 10.1097/00002030-199905280-00010. [DOI] [PubMed] [Google Scholar]
- 4.Badri M, Wood R. Usefulness of total lymphocyte count in monitoring highly active antiretroviral therapy in resource-limited settings. AIDS. 2003;17:541–545. doi: 10.1097/00002030-200303070-00009. [DOI] [PubMed] [Google Scholar]
- 5.Obirikorang C, Quaye L, Acheampong I. Total lymphocyte count as a surrogate marker for CD4 count in resource-limited settings. BMC Infect Dis. 2012;12:128–128. doi: 10.1186/1471-2334-12-128. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Kumarasamy N, Mahajan AP, Flanigan TP, Hemalatha R, Mayer KH, Carpenter CC, et al. Total lymphocyte count (TLC) is a useful tool for the timing of opportunistic infection prophylaxis in India and other resource-constrained countries. JAcquir Immune Defic Syndr. 2002;31:378–383. doi: 10.1097/00126334-200212010-00002. [DOI] [PubMed] [Google Scholar]
- 7.Spacek LA, Griswold M, Quinn TC, Moore RD. Total lymphocyte count and hemoglobin combined in an algorithm to initiate the use of highly active antiretroviral therapy in resource-limited settings. AIDS. 2003;17:1311–1317. doi: 10.1097/00002030-200306130-00005. [DOI] [PubMed] [Google Scholar]
- 8.Kamya MR, Semitala FC, Quinn TC, Ronald A, Njama-Meya D, Mayanja-Kizza H, et al. Total lymphocyte count of 1200 is not a sensitive predictor of CD4 lymphocyte count among patients with HIV disease in Kampala, Uganda. Afr Health Sci. 2004;4:94–101. [PMC free article] [PubMed] [Google Scholar]
- 9.Akinola NO, Olasode O, Adediran IA, Onayemi O, Murainah A, Irinoye O, et al. The search for a predictor of CD4 cell count continues: total lymphocyte count is not a substitute for CD4 cell count in the management of HIV-infected individuals in a resource-limited setting. Clin Infect Dis. 39:579–581. doi: 10.1086/422722. 200. [DOI] [PubMed] [Google Scholar]
- 10.Ledru E, Diagbouga S, Meda N, Sanou PT, Dahourou H, Ledru S, et al. A proposal for basic management of HIV disease in West Africa: use of clinical staging and haemogram data. Int J STD AIDS. 1998;9:463–470. doi: 10.1258/0956462981922638. [DOI] [PubMed] [Google Scholar]
- 11.Mahajan AP, Hogan JW, Snyder B, Kumarasamy N, Mehta K, Solomon S, et al. Changes in total lymphocyte count as a surrogate for changes in CD4 count following initiation of HAART: implications for monitoring in resource-limited settings. J Acquir Immune Defic Syndr. 2004;36:567–575. doi: 10.1097/00126334-200405010-00004. [DOI] [PubMed] [Google Scholar]
- 12.Renouf LS, Sheps S, Hubley A, Pick N, Johansen D, Tyndall MW. The role of diet in predicting iron deficiency anemia in HIV-positive women. Can J Diet Pract Res. 2012;73:128–133. doi: 10.3148/73.3.2012.128. [DOI] [PubMed] [Google Scholar]
- 13.Arnedo M, Alonso E, Eisenberg N, Ibáñez L, Ferreyra C, Jaén A, et al. Monitoring HIV viral load in resource limited settings: still a matter of debate? PLoS One. 2012;7: doi: 10.1371/journal.pone.0047391. [DOI] [PMC free article] [PubMed] [Google Scholar]