Impact of cytomegalovirus and grafts versus host disease on the dynamics of CD57+CD28-CD8+ T cells after bone marrow transplant

OBJECTIVES: The present study aimed to evaluate the dynamics of CD28 and CD57 expression in CD8+ T lymphocytes during cytomegalovirus viremia in bone marrow transplant recipients. METHODS: In a prospective study, blood samples were obtained once weekly once from 33 healthy volunteers and weekly from 33 patients. To evaluate the expression of CD57 and CD28 on CD8+ T lymphocytes, flow cytometry analysis was performed on blood samples for four months after bone marrow transplant, together with cytomegalovirus antigenemia assays. RESULTS: Compared to cytomegalovirus-seronegative healthy subjects, seropositive healthy subjects demonstrated a higher percentage of CD57+ and a lower percentage of CD28+ cells (p<0.05). A linear regression model demonstrated a continuous decrease in CD28+ expression and a continuous increase in CD57+ expression after bone marrow transplant. The occurrence of cytomegalovirus antigenemia was associated with a steep drop in the percentage of CD28+ cells (5.94 percent, p<0.01) and an increase in CD57+ lymphocytes (5.60 percent, p<0.01). This cytomegalovirus-dependent effect was for the most part concentrated in the allogeneic bone marrow transplant patients. The development of acute graft versus host disease, which occurred at an earlier time than antigenemia (day 26 vs. day 56 post- bone marrow transplant), also had an impact on the CD57+ subset, triggering an increase of 4.9 percent in CD57+ lymphocytes (p<0.05). CONCLUSION: We found continuous relative changes in the CD28+ and CD57+ subsets during the first 120 days post- bone marrow transplant, as part of immune system reconstitution and maturation. A clear correlation was observed between the expansion of the CD57+CD28-CD8+ T lymphocyte subpopulation and the occurrence of graft versus host disease and cytomegalovirus viremia.

Erros padröes mais úteis para fatores em modelos de dependência / More useful standard errors for factor effects in dependence models

Num contexto de modelos de dependência, tais como modelos lincares generalizados e modelos de taxas proporcionais, em que um fator explicativo seja de interesse, o cálculo correto da variância de funçöes dos estimadores correspondentes aos níveis do fator, tais como contrastes, näo é possível devido à omissäo das covariâncias estimadas. Sugerimos que os desvios-padröes sejam substituídos por quasi-desvios padröes, baseado em que aproximaçöes de variâncias de contrastes, por exemplo, podem ser calculadas. Mostramos como quasi-desvios padröes podem ser calculados, e identificamos classes de modelos lincares generalizados em que, sob certas condiçöes, as aproximaçöes de variâncias säo exatas para todos os contrastes pssíveis. Mostramos também como um intervalo para o erro cometido ao se usar essas aproximaçöes pode ser calculado, e sugerimos que tal intervalo seja também incluído na apresentaçäo dos resultados de ajuste do modelo