Preliminary Study Of Multi-Parameters In Predicting Response To Immunosuppressive Therapy In Patients With Aplastic Anemia

Objective To evaluate the predictive value of peripheral blood (PB) T lymphocyte subpopulation, bone marrow (BM) T lymphocytes intracellular IFN-γ, Th₁/Th₂ or Tc₁1/Tc₂ ratio and HLA-DRB1*1501 phenotype in predicting response to immunosuppressive therapy (IST) in patients with aplastic anemia (AA). To investigate the potential predictive value of combination of BM T lymphocytes intracellular IFN-γstain and HLA-DRB1*1501 detection in predicting response to IST in patients with AA. To establish a multivariable model of clinical, immunological and genetic parameters and discuss its potential possibility used as to predict responsiveness to IST in clinical practice in patients with AA.Methods Enrolled onto the present study were 51 primary acquired AA patients who were treated with cyclosporine A (CsA) based IST, consisting of 30 males and 21 females with median age of 32 (12～79y). Forty two (82.4%) cases were severe AA (SAA) and 9 (17.6%) cases were non-severe AA (NSAA) based on the classification criteria of AA. Eight (15.7%) patients were classified to very severe AA (VSAA) whose absolute neutriphil count (ANC) <0.2×10⁹/L. 39(68.6%) were previously treated patients and 12(23.5%) were newly diagnosed patients. 35(68.6%) patients were treated with CsA alone and the other 16(31.4%) patients were treated with antithymocyte globulin(ATG) or antilymphocyte globulin (ALG) plus CsA, no one was treated with ATG or ALG alone. In total patients, 35(68.6%) cases had a good response to therapy but the other 16(31.4%) were non-responders. All patients were serially analyzed for PB T lymphocytes subpopulation and bone marrow T cells intracellular IFN-γ, IL-4, Th₁/Th₂ and Tc₁/Tc₂ by flow cytometry (FCM) before and during treatment. The relationship between the laboratory indices and the clinical responsiveness are investigated and the potential predictive values of these immunological parameters in predicting response to IST in AA are evaluated and compared. HLA-DRB1*1501 phenotype were analyzed by PCR-SSP method using DNA extracted from PB before treatment in all the patients. The predictive potentials of different parameter combinations related to clinical responsiveness were statistically assessed. The study is aimed at establishing a multivariable model by using selected immunological and genetic parameters with high predictive value combined with clinical indices reflecting disease severity, and the possible utility of this model in predicting response to IST in patients with AA is then discussed.Resultsâ‘ Regarding the PB T lymphocyte subpopulation, it was found that the responders have a remarkable increased number in CD4⁺ cells as compared to the non-responders (P=0.003). PB T cell subpopulation was reexamined in 32 patients after IST and found that CD8⁺ cells were decreased while CD4⁺/CD8⁺ ratio was significantly increased (P=0.001 and 0.002) in responders. On the contrast, there were no differences in non-responders. No difference was found for responsive rate between the patients with a reverse CD4⁺/CD8⁺ ratio and those without a reverse CD4⁺/CD8⁺ ratio (54.5% vs 79.3% respectively).â‘¡In terms of BM T cell intracellular IFN-γ, patients with positive intracellular IFN-γhad a much higher responsive rate than those who were negative for intracellular IFN-γ[84.6%(33/39) vs 16.7%(2/12), P=0.001], which indicated that BM T cell intracellular IFN-γhad a potential predictive value. There was no difference in intracellular IFN-γstain reaction between CD4⁺ cells and CD8⁺ cells. In total patients, a positive BM T cell intracellular IFN-γstain had a sensitivity of 94.3% (33/35) and a specificity of 62.5% (10/16) to predict response to IST. The accuracy corresponding to clinical outcome was 84.3% (33+10/51). Its positive and negative predictive value was 84.6% (33/39) and 83.3% (10/12) respectively. On the other hand, in newly diagnosed patients, it had a high sensitivity of 100% (27/27) and a specificity of 58.3% (7/12) to predict response to therapy in the present study. The accuracy corresponding to clinical outcome was 87.2% (27+7/39) and its positive value was 84.4% (27/32).Furthermore, it had a high negative predictive value of 100% (7/7) in the patients studied. â‘¢ No remarkable differences were noticed in CD4⁺IFN- γ⁺, CD8⁺IFN- γ⁺, CD4⁺IL-4⁺ ,CD8⁺IL-4⁺, Th₁/Th₂ and Tc₁/Tc₂ between the responders and the non-responders in total patients. A negative BM T lymphocyte intracellular IFN-γ reaction strongly indicated non-responsiveness to IST because of its high sensitivity and negative predictive power. Furthermore, in patients positive for IFN-γ, the non-responders were demonstrated higher ratios of Th₁/Th₂ and Tc₁/Tc₂ than responders (P=0.043 and 0.041). In patients positive for intracellular IFN-γ stain, positive CD4⁺ cell intracellular IFN-γ and Th₁/Th₂<50 had a sensitivity of 54.5% (18/33) and a specificity of 66.7% (4/6), its positive and negative predictive value was 90.0% (18/20) and 21.1% (4/19) respectively, while positive CD8⁺ cells intracellular IFN-γ plus Tc₁/Tc₂<50 had a higher sensitivity of 72.7% (24/33) and a specificity of 66.7% (4/6), and its positive and negative predictive value was 92.3% (24/26) and 28.6% (4/14) respectively. It was implicated that a positive reaction of CD4⁺ or CD8⁺ lymphocytes intracellular IFN-γ combined with Th₁/Th₂<50 or Tc₁/Tc₂<50 may have a high predictive value of responsiveness to IST for patients with AA. Furthermore CD8⁺IFN-γ⁺ plus Tc₁/Tc₂<50 was superior to CD4⁺IFN-γ⁺ plus Th₁/Th₂<50. 28 patients were reexamined BM T cells" intracellular cytokines after treatment. It was found that IFN-γ was decreased (P=0.001) and IL-4 increased in both CD4⁺ and CD8⁺ cells (P=0.001 and 0.018), and both Th₁/Th₂ and Tc₁/Tc₂ were decreased (P=0.001) after treatment as compared to those before treatment in the responders, but there were no differences in non-responders. (3) A series of receiver operating characteristic (ROC) curves were used to compare the differences of the observed power among the parameters of PB CD3⁺, CD4⁺, CD8⁺ cells number, CD4⁺/CD8⁺ ratio, BM CD4⁺IFN-γ⁺, CD8⁺IFN-γ⁺ CD4⁺IL-4⁺ , CD8⁺IL-4⁺, Th₁/Th₂ and Tc₁/Tc₂, and it was shown that CD8⁺IFN-γ⁺ had the best observed power in prediction of response to therapy than the others. The area under the curve (AUC) of CD8⁺IFN-γ⁺ was 0.730 and had a remarkable statistical significance (P=0.009). When the cutoff point was selected as 8.90, this parameter had a sensitivity of 0.910 and 1-specificity (the possibility of making a false diagnose) of 0.500 to predicting responsiveness to therapy. (5) There were significant differences between the phenotype and genotype frequency of HLA-DRB1*1501 in 35 responders and 16 non-responders (0.457 vs 0.125, 0.263 vs 0.065 respectively, P<0.05). It was indicated that HLA-DRB1*1501 was of potential usefulness in predicting response to IST. The sensitivity and specificify of HLA-DRB1*1501 was 45.7% (16/35) and 87.5% (14/16) respectively. The diagnostic accuracy was 58.8% (16+14/51). The positive and negative predictive value was 88.9% (16/18) and 42.4% (14/33) respectively. With combination of intracellular IFN-γ with HLA-DRB1*1501, the serial test (both positive intracellular IFN-γ and HLA-DRB1 * 1501) showed a sensitivity of 42.9% (15/35) in predicting a good response to therapy, but its specificity was 93.7% (5+1+9/16) . The total diagnostic accuracy was 58.8% (15+5+1+9/51) .The positive and negative predictive value was 93.7% (15/16) and 42.8% .(5+1+9/23+2+10) respectively. After analysis of the ROC curves of intracellular IFN-γ, HLA-DRB1*1501 phenotype, the parallel test and serial test, it was found that the AUC of these four tests are 0.862, 0.872, 0.883, and 0.910 respectively, suggesting that the serial test had the largest AUC, and the combination of intracellular IFN-y and HLA-DRB1*1501 gene phenotype had the biggest observed power to predict the responsiveness. ?The predictive probabilities of 0.890, 0.774, 0.345 and 0.182 were calculated from a Logistic regression equation for the four groups of patients as follows: positive for both reaction, positive intracelluar IFN-γ and negative HLA-DRB1*1501, negative intracellular IFN-γ and positive HLA-DRB1*1501 and negative for both reaction, respectively. ?Based upon literature review and the data collected in the present study, ANC, CD8⁺ cells intracellular IFN-γ stain, Tc₁/Tc₂ ratio and HLA-DRB1*1501 phenotype were chosen to constitute a multivariable model consisting of clinical, immunological and genetic characteristics of the patients. An equation was obtained from the Logistic regression and a transient test variable containing predictive possibility of the above-mentioned four parameters was so achieved, then the transient variable was analyzed by ROC curve and a cutoff point was defined. This multivariable model and cutoff point can be used in predicting a given patient with AA to therapeutic response. Conclusionâ‘ This translational study investigated and evaluated certain predicts which seemed to be of potential predictive value in some pilot studies. It is confirmed that the immunological parameters indicating an abnormal immune function are essential to predict response to IST for idiopathic AA patients, but each of the immunological parameters used in the present study has different predictive power. Analysis of BM T cell intracellular IFN-γby FCM is of the greatest potentiality to be used in clinical practice because of its simplicity, practicability and superior sensitivity and specificity among all the parameters investigated in the present study , with the CD8⁺ intracellular IFN-γassay being the most powerful predict. It has a strong negative exclusive value for newly diagnosed patients in particular. However, its positive predictive value is relatively limited simply because the majority of idiopathic AA patients have a high positive rate of intracellular IFN-γstain. Although Th₁/Th₂ and Tc₁/Tc₂ ratios are not of much predictive value in predicting response to IST, if used lone, they do demonstrate an important predictive potentiality once combined with intracellular IFN-γreaction. The positive cellular IFN-γin CD8⁺ cells plus Tc₁/Tc₂<50 foresee a high possibility of achievement of remission after IST. Hence, the combination surpasses the intracellular IFN-γstain alone and provides a useful predictor to IST in AA patients with a positive stain. Although PB CD3⁺, CD4⁺, CD8⁺ cells and CD4⁺/CD8⁺ ratio are somewhat related to disease severity and treatment response, the lack of adequate sensitivity and specificity render them invalid for treatment prediction.â‘¡Combination of BM T cells intracellular IFN-γstain and HLA-DRB1*1501 gene phenotype can be a useful predictor for AA patients treated with IST. Superior to each of the single test, the combination shows a greatly improved sensitivity for either positive test in predicting response to IST because of inter-complementary effect from each test. The combination tests effectively increase the positive predictive value. Positive for both intracellular IFN-γstain and HLA-DRB1*1501 gene allele may predict a good response to therapy in the majority of AA patients. Patient either positive for both intracellular IFN-γreaction and HLA-DRB1*1501 phenotype or positive for intracellular IFN-γplus negative HLA-DRB1*1501 predicts a good outcome of treatment, with the former being of a superior predictive power over the latter. Above all, both tests are well established in methodology, simple and time-saving to be performed. Combination of the four tests provides physicians with a quick judgment for patient's possible response to treatment. (3) In the present study we attempt to establish a multivariable model intergrating clinical picture with immunological and genetic parameters to predict response to IST for AA patients and discuss its potential predictive value. The establishment of the model is greatly helpful to stratify the individual differences or heterogeneity of patients with AA in pathogenesis or genetic background in which patient may be unique to each other. It is indicated that the combination test is superior over each of the single test in terms of predictive power of therapeutic outcome. The higher accuracy and adapt to the complexity of the clinical situation inherited by the combination test make it potentially useful in clinical practice.