Reply to letter by Schwartz et al.

Schwartz et al. highlight the fact that several potential sources of bias may exist when estimating the cumulative incidence of brain metastases and corresponding incidence rates among patients with metastatic breast cancer (MBC). While it would be ideal to obtain patient-level data from all 25 articles included in this systematic review and meta-analysis, such data are not publicly available and are not feasible to obtain for all studies that were published over a 20-year time period. In the absence of patient-level data, total person-time can be estimated as (number of persons at risk) × (mean duration of time at risk per person). However, given that only one of the included studies reported mean follow-up duration, the median follow-up time was utilized instead. It is possible that using the median follow-up time may have introduced bias into the calculations, but the extent of bias (if any) and the direction of such bias cannot be determined given limitations in available data. As Schwartz et al. have indicated, the median follow-up in the included studies ranged from 12 to 137 months. We agree that this is problematic. Given that the incidence of brain metastases increases over time, higher incidence rates are generally reported in studies with longer follow-up durations.1 In fact, this was the motivation to perform a meta-analysis, which synthesizes available data and “corrects” for variable follow-up durations by estimating an incidence rate per patient-year. We acknowledge that several studies within our meta-analysis included patients who had brain metastases at initial diagnosis of MBC, which contribute to the total incidence of brain metastases during a patient’s disease course. Accordingly, several clinical trials evaluating the possible utility of brain metastases screening include MRI-based imaging of the brain both at initial diagnosis of MBC and longitudinally. When possible, we removed patients who had brain-only disease at the time of their MBC diagnosis, since the screening of the central nervous system (CNS) would not apply to this patient population. In cases when the presence or absence of concomitant extracranial metastases was unclear, we consistently omitted patients with brain metastases at initial diagnosis from our analysis (as Schwartz et al. suggested would be ideal). We repeated our analyses only omitting patients who had brain-only disease at presentation (including those with brain metastases and uncertain extracranial disease). Apart from a small increase in incidence of brain metastases per patient-year in the triple-negative MBC population from 0.13 in our published analysis2 to 0.14, the results were unchanged. This is because very few patients were reported to have CNS metastases at baseline. In addition to adjustments for baseline brain metastases, we also removed treatment arms that do not reflect standard clinical practice, such as the combination of neratinib plus paclitaxel3 and prophylactic cranial irradiation.4 Hence, the numbers that we used in our meta-analysis often differed from those reported within individual studies because we made every effort to account for potential sources of bias in this study. No specific funding sources were used for this work. Conflict of interest statement. K.J.J. reports the following: • Speaker/advisor board/consultant for: Amgen, Apobiologix, Eli Lilly, Eisai, Genomic Health, Pfizer, Roche, Novartis, Purdue Pharma. • Research funding: Eli Lilly, AstraZeneca. Authorship statement. Generation and drafting of manuscript: K.J.J. Conception and design of experiment: M.K., Y.G., A.K., and K.J.J. Data collection: M.K. and K.J.J. Data analysis and interpretation: M.K., Y.G., A.K., and K.J.J. Manuscript revision and final approval: M.K., Y.G., A.K., and K.J.J.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.