The association of P‐glycoprotein with response to chemotherapy and clinical outcome in patients with osteosarcoma

Abstract

BACKGROUND There is controversy regarding whether P‐glycoprotein (Pgp) may be a prognostic factor for the response to chemotherapy and clinical disease progression in patients with osteosarcoma. METHODS The authors conducted a meta‐analysis of 14 studies (n = 631 patients) that evaluated the correlation between Pgp and histologic response to chemotherapy and clinical disease progression (death, metastasis, or recurrence). Data were synthesized in receiver operating characteristic curves and with fixed‐effects and random‐effects likelihood ratios and risk ratios. RESULTS Pgp had no discriminating ability for identifying poor responders versus good responders to chemotherapy: The positive likelihood ratio was 1.15 (95% confidence interval [95% CI], 0.93–1.43), and the negative likelihood ratio was 0.88 (95% CI, 0.65–1.18; random‐effects calculations). There was some between‐study heterogeneity, but no study showed strong discriminating ability. Conversely, Pgp positivity increased the risk of disease progression 1.92‐fold (95% CI, 1.18–3.13; random‐effects calculations) with some between‐study heterogeneity that disappeared when only studies that employed immunohistochemistry were considered (risk ratio, 2.23; 95% CI, 1.37–3.64). The results were robust in various sensitivity analyses, although smaller studies tended to show stronger associations with the risk of disease progression compared with larger studies (P = 0.03). CONCLUSIONS The available evidence showed conclusively that Pgp was not associated with the histologic response of patients with osteosarcoma to combination chemotherapy regimens. Conversely, Pgp positivity, as determined by immunohistochemistry, was a strong correlate of more rapid disease progression, although there was heterogeneity across the performed studies that, to some extent, may have reflected bias, differential measurements of Pgp, or confounding with other risk factors. Cancer 2003;98:581–9. © 2003 American Cancer Society. DOI 10.1002/cncr.11546