Clinical evaluation of lapatinib therapy in metastatic breast cancer using the bayes meta-analysis

Tadao Inoue, Tomokazu Iyoda, Wataru Yamamoto, Yoshio Uetsuka

Research output: Contribution to journalReview articlepeer-review


The efficacy of treatments involving lapatinib for patients with metastatic breast cancers was evaluated in a Bayesian meta-analysis of published data from randomized controlled clinical trials. Four randomized controlled trials including 2,708 patients met the inclusion criteria. Among these patients, 568 were positive for the human epidermal growth factor receptor 2 (HER2). The clinical benefit rate (CBR) for HER2-positive patients was the primary outcome of the analysis, and the overall survival (OS) and the number needed to treat (NNT) were the secondary outcomes of the reported meta-analysis. The Bayesian meta-analysis was conducted according to the Markov-chain Monte-Carlo technique in WinBUGS. The CBR for HER2-positive patients was significantly improved (odds ratio [OR]: 2.281, 95% confidence interval [CI]: 1.490-3.628), whereas no statistically significant improvement was seen in the overall patient CBR (OR: 1.559, 95% CI: 0.768-3.238). The OS hazard ratio (HR) and NNT for the CBR were also estimated for HER2-positive patients. The difference in the OS HR was not statistically significant (HR: 0.789, 95% CI: 0.556-1.086) for HER2-positive patients. The improvement in the NNT for the CBR was statistically significant (NNT 5.164, 95% CI: 3.803-8.723) for HER2-positive patients.

Original languageEnglish
Pages (from-to)347-352
Number of pages6
JournalJapanese Journal of Cancer and Chemotherapy
Issue number3
Publication statusPublished - 2014 Mar
Externally publishedYes


  • Bayesian meta-analysis
  • Lapatinib
  • Markov-chain Monte-Carlo methods
  • Metastatic breast cancer

ASJC Scopus subject areas

  • Medicine(all)


Dive into the research topics of 'Clinical evaluation of lapatinib therapy in metastatic breast cancer using the bayes meta-analysis'. Together they form a unique fingerprint.

Cite this