A Study of Alternate Biomarkers in HIV Disease and Evaluating their Efficacy in Predicting T CD4+ Cell Counts and Disease Progression in Resource Poor Settings in Highly Active Antiretroviral Therapy (HAART) Era
K V Ramana1, V Sabitha2, Ratna Rao3
1 Department of Microbiology, Apollo Health City, Jubilee Hills, Hyderabad, Andhra Pradesh, India.
2 Department of Biochemistry, Chalmeda Anandarao Institute of Medical Sciences, Bommakal, Karimnagar, Andhra Pradesh, India.
3 Department of Microbiology, Apollo Health City, Jubilee Hills, Hyderabad, Andhra Pradesh, India.
Name, Address, E-Mail Id of The Corresponding Author: Dr K V Ramana, Department of Microbiology, Apollo Health City, Jubilee Hills, Hyderabad, Andhrapradesh, India.
Phone: 9440704234
E-mail: ramana_20021@rediffmail.com
Introduction: Human Immunodeficiency Virus (HIV), the causative agent of AIDS, has been a challenge to medical fraternity since it was first discovered in 1983. About 40 million people are living with HIV infection globally and 99% of the infected people are in south East Asia (SEA). Traditionally, HIV disease and progression, initiation of HAART and response to therapy is monitored by assessing in regular intervals, the T CD4+ cell counts and plasma HIV/RNA viral load. Resource poor, low and low – middle income group countries still have no finances to acquire infrastructure and scientific technology for performing such tests.
Objectives: Since very few studies are available, they have demonstrated the role of alternate biomarkers that can be used to predict CD4 cell counts and thereby, monitor HIV disease progression and HAART. We aimed to measure certain haematological parameters in HIV seropositive patients and to evaluate their efficacy in predicting TCD4+ cell counts.
Methods: The study group included 250 HIV seropositive patients with an age range of 18-65 years. 140(56%) males and 110(44%) females were included in the study. Absolute TCD4+cell counts and CD8+T cell counts were measured by using a flow cytometer. (MMWR Recommendations and Reports, 1992) TLC; HB%, AEC and ESR were estimated by using conventional haematological methods. CRP was evaluated by latex agglutination test (Immuno CRP Latex Agglutination Test).
Results: Among the tested haematological markers, a TLC of <1800 cells/mm3 showed high specificity (100%) in predicting CD4 counts of < 200 cells/mm3, with an accuracy of 61.46%. Haemoglobin and Absolute Eosinophilic counts showed high specificities of 84.09% and 94.32% respectively in predicting CD4 counts which were below 350 cells/mm3. ESR with 98.98% sensitivity and AEC which had 83.67% sensitivity were able to predict CD4 counts of <200 cells/mm3.
Conclusion: Among the tested biomarkers, it was seen that Absolute Eosinophilic counts of more than 550 cells/mm3, Blood Haemoglobin which was less than 10 g%, ESR which measured more than 20 mm, CRP values of >1.2 and TLC of <1800 cells/mm3 could be helpful in predicting CD4 cell counts of < 350 and <200 cells/mm3.
Introduction
Human Immunodeficiency Virus (HIV), the causative agent of AIDS, has been a challenge to medical fraternity since it was first discovered in 1983 [1]. Laboratory diagnosis is affected by the window period in which patients show non – specific symptoms and the serum anti – HIV antibodies are present in undetectable amounts. Though the availability of advanced third and fourth generation tests have solved the problem, the cause of concern is the laboratory monitoring of the infected individuals. About 40 million people are living with HIV infection globally and 99% of the infected people are in south East Asia (SEA) [2]. With inception of Highly Active Antiretroviral Therapy (HAART), the quality of life of HIV infected individuals is gradually improving [3]. The number of people which contract new infections has been on decline globally and those who have access to HAART are increasing in number. Antiretroviral therapy was made available in India since April 2004 and currently, over three lakh people are on HAART [4]. The cause of concern now is the availability and efficacy of cost-effective disease monitoring strategies. T CD4+ cell counts and serum/plasma HIV/RNA viral load, which are traditional biomarkers for assessing the disease progression and for monitoring HAART, require sophisticated equipment and they cannot be affordable to financially constrained, low middle and low income group countries which actually carry a majority of burden of HIV seropositive individuals. We attempted to evaluate various haematological parameters in HIV seropositive patients and to evaluate their efficacy in predicting T CD4+ cell counts.
Material and Methods
This study was carried out at Apollo Health City, a tertiary care hospital, between June 2011 to May 2012, which included 250 HIV sero-positive and antiretroviral therapy naive individuals. HIV sero – positive patients who attended Integrated Counseling and Testing Centre (ICTC) which was situated at Area Hospital, Siddipet, India, were enrolled in the study. The study groups, at the time of their enrollment, were antiretroviral therapy naive, and those who were presently on HAART were excluded from the study. All the subjects who were included in the study were provided with a proforma with details of the study and informed and written consents were obtained from them. The study was approved by institutional ethical board. Blood samples were collected as per standard laboratory procedures. The HIV statuses of the subjects were confirmed as per NACO guidelines by using three different ELISA methods [5]. Absolute TCD4+cell counts and CD8+T cell counts were measured by using a flow cytometer. (MMWR Recommendations and Reports, 1992) TLC; HB%, AEC and ESR were estimated by using conventional haematological methods. CRP was evaluated by latex agglutination test (Immuno CRP Latex Agglutination Test).
Statistical software: Descriptive statistical analysis was carried out in the present study. The statistical softwares, namely, SAS 9.2, SPSS 15.0, Stata 10.1, MedCalc 9.0.1, Systat 12.0 and R environment ver. 2.11.1 were used for the analysis of the data (Spearman’s rank order correlation, Sensitivity, specificity, Positive predictive value (PPV), Negative predictive value (NPV) Accuracy and p – value). Microsoft Word and Excel were used to generate tables.
Results
The median age of the subjects who were included in the study was 33 years, with an age range of 18-65 years. 140(56%) males and 110 (44%) females were included in the study. The cut – off for T CD4 + cell counts was taken as < 200 cell/mm3 and < 350 cell/mm3. The other variables with different cut-off values, which included ESR (>20 mm), AEC (>350 and > 550 cell/mm3), TLC (< 2500 cell/mm3 and <1800 cell/mm3), HB (<10) and CRP (>1.2) were included, based on the WHO criteria and the published literature. The diagnostic performance of each variable in predicting T CD4 + cell counts, at a cut-off of < 200 cell/mm3, has been shown in [Table/Fig-1]. Performance of each variable at its respective cut-off, in predicting T CD4 + cell counts of < 350 cell/mm3, has been shown in [Table/Fig-2]. Spearman’s rank order correlation and p-value of each variable at their cut-off values with respect to predicting T CD4 + cell counts (< 200, 200-350, > 350, 200-500 and > 500 cell/mm3) have been shown in [Table/Fig-3].
Diagnostics performance of ESR, AEC, TLC, Hb and CRP for predicting the low CD4 count (<200)
| Sensitivity | Specificity | PPV | NPV | Accuracy | p value |
|---|
| ESR>20 | 98.98 | 15.56 | 43.56 | 96.15 | 49.00 | <0.001* |
| AEC>350 | 83.67 | 58.94 | 56.94 | 84.76 | 68.67 | <0.001* |
| AEC >550 | 30.61 | 90.07 | 66.67 | 66.67 | 66.67 | <0.001* |
| TLC <2500 | 31.63 | 88.08 | 63.27 | 66.50 | 65.88 | <0.001* |
| TLC <1800 | 2.00 | 100.00 | 100.00 | 61.13 | 61.46 | <0.001* |
| Hb<10 | 37.76 | 77.48 | 52.11 | 65.73 | 61.85 | 0.009* |
| CRP>1.2 | 59.18 | 68.87 | 55.24 | 72.22 | 65.06 | <0.001* |
*p Value statistically significant
Diagnostics performance of ESR, AEC, TLC, Hb and CRP for predicting the low CD4 count (<350)
| Sensitivity | Specificity | PPV | NPV | Accuracy | p value |
|---|
| ESR>20 | 94.41 | 19.32 | 68.16 | 65.38 | 67.87 | <0.001* |
| AEC>350 | 70.19 | 64.77 | 78.47 | 54.28 | 68.27 | <0.001* |
| AEC >550 | 24.84 | 94.32 | 88.89 | 40.69 | 49.40 | <0.001* |
| TLC <2500 | 26.09 | 92.05 | 85.71 | 40.50 | 49.40 | <0.001* |
| TLC <1800 | 1.24 | 100.00 | 100.00 | 35.63 | 36.14 | 0.541 |
| Hb<10 | 35.40 | 84.09 | 80.28 | 41.57 | 52.61 | 0.001* |
| CRP>1.2 | 47.20 | 67.05 | 72.38 | 40.97 | 54.22 | 0.032* |
*p Value statistically significant
Spearman Rank correlation CD4 with AEC, TLC, HB, ESR and CRP
| Pair | CD4 count |
|---|
| <200 | 200-350 | <350 | 200-500 | >500 | Total |
|---|
| r | p | r | p | r | p | r | p | r | p | r | p |
|---|
| CD4 vs AEC | -0.115 | 0.26 | -0.155 | 0.224 | -0.084 | 0.439 | -0.38 | <0.001* | -0.341 | 0.065 | -0.444 | <0.001* |
| CD4 vs TLC | 0.086 | 0.403 | 0.199 | 0.117 | 0.264 | 0.013* | 0.304 | <0.001* | 0.14 | 0.462 | 0.381 | <0.001* |
| CD4 vs Hb | 0.377 | <0.001* | -0.013 | 0.92 | 0.04 | 0.714 | 0.339 | <0.001* | 0.087 | 0.648 | 0.329 | <0.001* |
| CD4 vs ESR | -0.108 | 0.291 | -0.029 | 0.82 | -0.1 | 0.359 | -0.344 | <0.001* | -0.2 | 0.289 | -0.376 | <0.001* |
| CD4 vs CRP | -0.299 | 0.003* | -0.222 | 0.080+ | -0.138 | 0.198 | -0.33 | <0.001* | -0.204 | 0.279 | -0.341 | <0.001* |
Classification of Correlation Co-efficient (r)
Up to 0.1 : Trivial Correlation; 0.1-0.3 : Small Correlation; 0.3-0.5 : Moderate Correlation; 0.5-0.7 : Large Correlation
0.7-0.9 : V.Large Correlation; 0.9- 1.0 : Nearly Perfect correlation; 1 : Perfect correlation; *P Value Statistically significant
Discussion
Infection with HIV results in a variable disease progression, where many HIV infected individuals survive for more than 5 years even without undergoing an effective antiretroviral therapy [6]. A close monitoring of HIV infected patients is essential, for reducing their morbidity and mortality. Traditionally, HIV disease and progression, initiation of HAART and response to therapy were monitored by assessing in regular intervals (6 months), the TCD4+ cell counts and plasma HIV/RNA viral load. Resource poor, low and low middle income group countries still have no finances to acquire infrastructure and scientific technology for performing such tests. Very few studies are available, that have demonstrated the role of alternate biomarkers that could be used to predict CD4 cell counts and thereby, monitor HIV disease progression and HAART. The present study used CD4 cell cutoffs of < 200 cells/mm3 and < 350 cells/mm3 and it compared the usefulness of alternate biomarkers. Among the tested alternative biological markers, a TLC of <1800 cells/mm3 showed high specificity (100%) in predicting CD4 counts of < 200 cells/mm3, with an accuracy of 61.46%. Haemoglobin and Absolute Eosinophilic counts showed high specificities of 84.09% and 94.32% respectively in predicting CD4 counts which were below 350 cells/mm3. ESR with 98.98% sensitivity and AEC which had 83.67% sensitivity were able to predict CD4 counts of <200 cells/mm3. Absolute Eosinophilic counts of more than 550 cells/mm3 showed 90.07 specificity and a CRP of >1.2 mg/dL showed 68.87 specificity in predicting CD4 counts of <200 cells/mm3. In agreement with other recent studies, our studies also suggested that TLCs of < 1800 and < 2000 cells/mm3 may be more suitable for initiation of HAART. A moderate correlation was observed for Absolute Eosinophilic counts (r -0.444 and p <0.001) in predicting CD4 counts of <500 cells/mm3. Haemoglobin (r 0.377 p <0.001) was a useful alternative for predicting CD4 counts of <200 cells/mm3. ESR (r -0.376 p <0.001) and CRP (r -0.341 p <0.001) also correlated well with CD4 cell counts. As was observed by other studies, a TLC of <1200 cells/mm3 was ineffective in predicting T CD4+ cell counts. We tried values of TLC at two different cutoffs of <1800 and <2500 cells/mm3. A recent study tried to predict T CD4 cell counts of <200 cells/mm3 by using the WHO HIV/AIDS clinical staging and found it to have a low sensitivity [7,8]. An Indian study evaluated haemoglobin, ESR and TLC and found that a TLC of < 1200 cells/mm3 could predict CD4 counts of <200 cells/mm3, with very low sensitivity [9]. Azzoni et al., recently used a prediction based classification(PBC) method by using alternate biomarkers like total lymphocyte count, in predicting CD4 cell counts, thereby reducing the cost of patient management [10]. Boulware et al studied some biomarkers and found that a pre HAART evaluation of such markers could predict Immune Reconstitution Inflammatory Syndrome (IRIS) [11,12]. Another study which was done in a developing country, tried to evaluate non physician clinicians’ roles in staging and monitoring HIV diastase [13]. Absolute Eosinophilic counts correlated well with CD4 cell counts in predicting the disease stage and thereby, they may prove to be a useful biomarker in assessing the disease progression, as was observed by Cohen et al., [14]. Assessment of utility of ESR in the current study revealed that a raised ESR correlated well with CD4 cell counts (Accuracy 67.87%) and that it could predict CD4 counts which were below 350 cells/mm3. It can be used in decision making, before starting on HAART and thereby for monitoring antiretroviral therapy [15]. Few previous studies found no correlation between TCD4+ cell counts of less than 200 cells/mm3 and ESR [16]. Only scanty data on the usefulness of ESR in predicting CD4 cell counts was found to be available on reviewing the literature [15,17,18]. Doing several studies on the utility of alternate biological markers in HIV could pave the way for a better understanding of immunopathogenesis of HIV disease and for improving laboratory testing strategies that can be used in predicting clinical stage, initiation of HAART and management of HIV disease [4]. Current study’s results correlated well with those of a recent study, which found anaemia to be a useful predictor of CD4 cell counts of < 200 cells/mm3 [19]. An Indonesian study showed good correlation of anaemia with reduced CD4 cell counts, as was seen in the current study, which observed that anaemia showed high specificity (77.48%) and accuracy in predicting CD4 cell counts of < 200 cells/mm3. Elevated CRP was a good predictor of CD4 counts of <200 cells/mm3, as was observed in the current study [20]. A moderate correlation was observed in our study between anaemia and CD4 cells of <200 cells/mm3, as was observed in the study of Mata-Marin et al., [21]. Suresh Venkata Satya Attili et al., in a study which was done in northern India, studied the haematological profiles of 470 HIV seropositive patients and found that moderate to severe anaemia and leucopaenia were associated with CD4 cell counts of < 200 cells/mm3 [22]. Haematological abnormalities are a cause of concern in HIV infected individuals and they could be useful in management of disease. Very few studies are available from India in this regards. This study also re emphasised the need of rethinking over a TLC of <1200 cells/mm3 in predicting CD4 cell counts of <200 cell/mm3 and initiation of HAART [7,9]. In another study which was done in Uganda, Miiro et al., evaluated low cost alternate blood markers in predicting CD4 counts of < 200 cell/mm3and found that anaemia had a sensitivity (95% CI) of 47% and a specificity of 76% [23]. An Iranian study evaluated TLC counts of <1300 cells/mm3 (which were slightly higher than the WHO recommended values) and found them to have optimum sensitivity and specificity in predicting CD4 counts of < 200 cell/mm3 [24]. Marrigje A. de Jong et al., from Indonesia studied the utility of TLC among those who were on HAART and found that it was effective in predicting CD4 counts of <200 cells/mm3 [25]. The study of Helena P W et al., revealed that a TLC of <1700 cells/mm3 could predict CD4 counts of <350 cells/mm3 [26].
Conclusion
The current study’s results clearly indicated abnormal haematological parameters in HIV sero-positive patients. Among the tested biomarkers, it was seen that Absolute Eosinophilic counts of more than 550 cells/mm3, Blood Haemoglobin of less than 10 g%, ESR which was more than 20 mm, a CRP value of >1.2 and a TLC of <1800 cells/mm3 could be helpful in predicting CD4 cell counts of < 350 and <200 cells/mm3. Such alternate biological blood parameters can be regularly measured in HIV sero-positive individuals, that are cost effective in making decisions on the start of HAART and they can be used further in the management
of HIV disease and treatment response.
*p Value statistically significant*p Value statistically significantClassification of Correlation Co-efficient (r)Up to 0.1 : Trivial Correlation; 0.1-0.3 : Small Correlation; 0.3-0.5 : Moderate Correlation; 0.5-0.7 : Large Correlation0.7-0.9 : V.Large Correlation; 0.9- 1.0 : Nearly Perfect correlation; 1 : Perfect correlation; *P Value Statistically significant
[1]. Barré-Sinoussi F, Chermann JC, Rey F, Nugeyre MT, Chamaret S, Gruest J, Isolation of a T-lymphotropic retrovirus from a patient at risk for AIDS Science 1983 220:868-71. [Google Scholar]
[2]. UNAIDS Epidemic update 2007. http://data.unaids.org/pub/PressRelease/2007/070706_indiapressrelease_en.pdf [Google Scholar]
[3]. Palella FJ Jr, Delaney KM, Moorman AC, Loveless MO, Fuhrer J, Satten GA, Declining morbidity and mortality among patients with advanced HIV infection. HIV outpatient study investigators N Engl J Med 1998 338:853-60. [Google Scholar]
[4]. Vajpayee M, Mohan T, Current practices in laboratory monitoring of HIV infection Indian J Med Res 2011 134(6):801-22.doi: 10.4103/0971-5916.92627 [Google Scholar]
[5]. NACOHIV/AIDS Epidemiological Surveillance and Estimation Report for the Year 2005. New Delhi National AIDS Control Organisation 2006 http://www.aidsportal.org/repos/fnlapil06rprt.pdf [Google Scholar]
[6]. Ramana KV, Jagadeeswhwara chary, Sabitha V, Mohanty SK, Ratna Rao, Role of Hematological and Alternate Markers in Human Immunodeficiency Virus Disease Progression Am Med J 2010 1(2):84-87. [Google Scholar]
[7]. Baveewo S, Ssali F, Karamagi C, Kalyango JN, Hahn JA, Ekoru K, Validation of World Health Organisation HIV/AIDS Clinical Staging in Predicting Initiation of Antiretroviral Therapy and Clinical Predictors of Low CD4 Cell Count in Uganda PLoS One 2011 May 12 6(5):e19089doi: 10.1371/journal.pone.0019089 [Google Scholar]
[8]. WHO: Scaling up antiretroviral therapy in rei-limited settings: treatment guidelines for a public health approach – 2003 revision (Monograph online). Switzerland: WHO press; 2004. Available from: http://www.who.inUhiv/pub/prev 3areieniarvrevision2003en.pdf [Google Scholar]
[9]. Sen Sourav, Vyas Akshat, Sanghi Sunil, Shanmuganandan K, Gupta RM, Kapila Ketoki, Correlation of CD4+ T cell Count with Total Lymphocyte Count, Haemoglobin and Erythrocyte Sedimentation Rate Levels in Human Immunodeficiency Virus Type-l Disease MJAFI 2011 67:15-20. [Google Scholar]
[10]. Azzoni L, Foulkes AS, Liu Y, Li X, Johnson M, Smith C, Prioritizing CD4 Count Monitoring in Response to ART in Rei-Constrained Settings: A Retrospective Application of Prediction-Based Classification PLoS Med 2012 9(4):e1001207doi:10.1371/journal.pmed.1001207 [Google Scholar]
[11]. Boulware DR, Meya DB, Bergemann TL, Wiesner DL, Rhein J, Musubire A, Clinical Features and Serum Biomarkers in HIV Immune Reconstitution Inflammatory Syndrome after Cryptococcal Meningitis: A Prospective Cohort Study PLoS Med 2010 7(12):e1000384doi:10.1371/journal.pmed.1000384 [Google Scholar]
[12]. French MA, Immune reconstitution inflammatory syndrome: immune restoration disease 20 years on Med J Aust 2012 Mar 19 196(5):318-21. [Google Scholar]
[13]. Brentlinger PE, Torres JV, Martínez PM, Ghee A, Lujan J, Bastos R, Clinical Staging of HIV-Related Illness in Mozambique: Performance of Non physician Clinicians Based on Direct Observation of Clinical Care and Implications for Health Worker Training J Acquir Immune Defic Syndr 2010 55:351-55. [Google Scholar]
[14]. Cohen AJ, Steigbigel RT, Eosinophilia in Patients Infected with Human Immunodeficiency Virus J Infect Dis 1996 174:615-18. [Google Scholar]
[15]. Ndakotsu MA, Salawu L, Durosinmi MA, Relation between erythrocyte sedimentation rate, clinical and immune status in HIV-infected patients Niger J Med 2009 18(2):208-10.[PubMed: 19630332] [Google Scholar]
[16]. Morpeth SC, Crump JA, Shao HJ, Ramadhani HO, Kisenge PR, Moylan CA, Predicting CD4 lymphocyte count <200 cells/mm(3) in an HIV type 1-infected African population AIDS Res Hum Retroviruses 2007 23(10):1230-6.[PubMed: 17961109] [Google Scholar]
[17]. Lowe DM, The Erythrocyte sedimentation rate in HIV: a neglected Parameter? AIDS 2010 Nov 27 24(18):2773-5.doi: 10.1097/QAD.0b013e32833ef5d7 [Google Scholar]
[18]. Peters PJ, Zulu I, Kancheya NG, Lakhi S, Chomba E, Vwalika C, Rwanda Zambia HIV Research Group. Modified Kigali combined staging predicts risk of mortality in HIV-infected adults in Lusaka, Zambia AIDS Res Hum Retroviruses 2008 24(7):919-24.[PubMed: 18593343] [Google Scholar]
[19]. Owiredu WK, Quaye L, Amidu N, Addai-Mensah O, Prevalence of anaemia and immunological markers among Ghanaian HAART-naïve HIV-patients and those on HAART Afr Health Sci 2011 Mar 11(1):2-15. [Google Scholar]
[20]. Wisaksana R, Sumantri R, Indrati AR, Zwitser A, Jusuf H, de Mast Q, Anemia and iron homeostasis in a cohort of HIV-infected patients in Indonesia BMC Infect Dis 2011 11:213 [Google Scholar]
[21]. Mata-Marín JA, Gaytán-Martínez JE, Martínez-Martínez RE, Arroyo-Anduiza CI, Fuentes-Allen JL, Casarrubias-Ramirez M, Risk factors and correlates for anemia in HIV treatment-naïve infected patients: a cross-sectional analytical study BMC Res Notes 2010 Aug 20 3:230doi: 10.1186/1756-0500-3-230 [Google Scholar]
[22]. Attili Suresh Venkata Satya, Singh VP, Rai Madhukar, Varma Datla Vivekananda, Gulati K, Sundar Shyam, Hematological profile of HIV patients in relation to immune status – a hospital-based cohort from Varanasi, North India Turk J Hematol 2008 25:13-19. [Google Scholar]
[23]. Miiro G, Nakubulwa S, Watera C, Munderi P, Floyd S, Grosskurth H, Evaluation of affordable screening markers to detect CD4+ T cell counts below 200 cells/μl among HIV-1-infected Ugandan adults Trop Med Int Health 2010 Apr 15(4):396-404.doi: 10.1111/j.1365-3156.2010.02471.x. Epub 2010 Feb 17 [Google Scholar]
[24]. Jafari S, Rasoolinejad M, Kouchak Emadi H, Mokarami F, The relationship between total lymphocyte count and CD4 count in patients infected with HIV Tehran Medical Journal 2009 67(7):477-82. [Google Scholar]
[25]. de Jong MA, Wisaksana R, Meijerink H, Indrati A, van de Ven AJ, Alisjahbana B, Total lymphocyte count is a reliable surrogate marker for CD4 cell counts after the first year of antiretroviral therapy: data from an Indonesian cohort study Trop Med Int Health 2012 May 17(5):581-3.doi: 10.1111/j.1365-3156.2012.02961.x. Epub 2012 Feb 26 [Google Scholar]
[26]. Oudenhoven HP, Meijerink H, Wisaksana R, Oetojo S, Indrati A, van der Ven AJ, Total lymphocyte count is a good marker for HIV-related mortality and can be used as a tool for starting HIV treatment in a rei-limited setting Trop Med Int Health 2011 Nov 16(11):1372-9.doi: 10.1111/j.1365-3156.2011.02870.x. Epub 2011 Sep 1 [Google Scholar]