Search strategy

Based on the PRISMA-IDP Statement,18 electronic searches were performed in PubMed, Embase, Web of Science, and Cochrane Library using the following retrieval strategy: ((HER2 OR (Human epidermal growth factor receptor 2)) AND ((“Breast Neoplasms”[Mesh]) OR (Breast cancer) OR (Breast tumour) OR (Breast tumour) OR (Breast carcinoma) OR (Breast Neoplasm))) AND ((((Neoadjuvant therapy) OR (Neoadjuvant treatment)) AND ((Adjuvant therapy) OR (Adjuvant treatment)) AND (Trastuzumab OR Herceptin) AND (Overall Survival)) OR (Mastectomy AND ((Breast conservation surgery) OR (Breast conserving surgery) OR (Breast preservative surgery) OR (Breast preservation surgery) OR (Breast preserving surgery) OR (Breast preservative surgery) OR (Breast conservation therapy) OR (Breast conserving therapy) OR (Breast preservative therapy) OR (Breast preservation therapy) OR (Breast preserving therapy) OR (Breast preservative therapy) OR (Breast conservation treatment) OR (Breast conserving treatment) OR (Breast preservative treatment) OR (Breast preservation treatment) OR (Breast preserving treatment) OR (Breast preservative treatment)) AND ((Local-regional relapse) OR (Local-regional recurrence) OR (Local relapse) OR (Local recurrence)))). No restrictions were required during the retrieval. The searching of citation was terminated as of 28th May 2019.

Inclusion criteria

  • Early HER2-enriched breast cancer patients;
  • English publications which covered the number or the rate of event and the total sample size;
  • Clinical trials which concurrently documented the data of LRR after BCT and that after mastectomy or OS after NAT and that after AT. LRR referred to the first recurrence of tumor in the ipsilateral breast, the chest wall or regional lymphatics, without evidence of distant metastasis. The definition of BCT was a combination of radiotherapy and surgery strategy involving lumpectomy, segmental mastectomy, quadrantectomy, or wedge resection. The methods of mastectomy included simple mastectomy and modified radical mastectomy. OS ranged from the date of diagnosis to the date of death or loss to follow-up.

Exclusion criteria

  • Trials published in non–English;
  • Patients with moderate-to-advanced stage disease;
  • Male patients;
  • Reviews, case reports, and conference papers;
  • Other details that did not meet the inclusion criteria.

Two co-authors (Qian Wu and Jing Xiong) independently screened the retrieved citations and reserved the pertinent studies according to the titles, abstracts, and full-text articles. If some disagreements surfaced, they were resolved by the third reviewer (Zhumin Su).

Data abstraction

The following information was abstracted using Microsoft Excel version 2016 (Microsoft Corporation, Redmond, Washington, USA) by two authors (Jing Xiong and Zhumin Su): first author, publication year, study duration, original nation, median age, median follow-up, whether patients received trastuzumab treatment, the number of patients undergoing BCT or mastectomy and how many of them developed LRR, and the number of women who received NAT or AT and how many died from either treatment paradigms. Provided that there were any inconsistencies, they were resolved by discussion.

Related Articles

Statistical analysis

The included clinical studies were discussed by their differences of eligible criteria, chemotherapy strategy, and implementation of radiotherapy, as well as the risk of bias of study evaluation using the new version of the Cochrane tool. The crude odds ratio (OR) with its 95% confidence interval (CI) of LRR with respect to mastectomy compared to BCT or OS regarding NAT versus AT in each trial was calculated and then pooled together. If the event number was not shown in the publication, it was computed based on the end point percentages of LRR and OS or other information. A heterogeneity Chi2 test with a significance level of P<0.1 was used to estimate the heterogeneity among different studies;19 when it was not significant (P>0.1), a fixed-effect Mantel-Haenszel model was utilized to pool the data, otherwise, a random-effect Mantel-Haenszel model was applied.19 The publication bias was assessed by creating a Begg’s funnel plot with 95% CI and Egg’s test with a significance level of P<0.05. All statistical tests were performed in StataSE version 12.0 (Stata Corporation, College Station, TX, USA).

READ FULL ARTICLE Curated publisher From Dovepress