Dear Editor(s),

We thank you for the opportunity to respond to the letter by Schöpfer et al. regarding our recent meta-analysis published in Psychological Medicine (Cheng et al., Psychological Medicine, Reference Cheng, Tsang, Han, Zhang, Maes, Klugah-Brown, Bore and Becker2026;56:e1). We are grateful for their thoughtful comments, which prompted us to clarify methodological decisions and conduct additional analyses. The supplementary analyses generally corroborate the central finding of our work, suggesting altered dorsal anterior cingulate cortex (dACC) activation – but not amygdala or hippocampal alterations – in depression during autobiographical memory recall.

Schöpfer et al. correctly note that five studies used designs with more than two groups and reported omnibus F-tests without the post-hoc pairwise statistics required for seed-based d mapping (SDM). Because a principal aim of our meta-analysis was to maximize the inclusion of relevant data, we adopted a pragmatic approach in these cases, extracting the depression-versus-healthy-control contrast that was most closely aligned with our meta-analytic focus. We acknowledge that this conversion, and the selection of particular contrasts, may introduce bias. Moreover, inconsistencies in the direction of effects – for example, between individuals at risk for depression and patients with depression – cannot be fully modeled within the coordinate-based framework applied here. To address this concern directly, we conducted a sensitivity analysis excluding the affected studies. The right paracingulate cortex (dorsal anterior cingulate cortex, dACC), which formed the largest and most robust cluster in the primary analysis, remained significant (SDM-Z value = 3.315; p = .000458, voxels = 34), although its spatial extent was reduced, as expected given the loss of statistical power. However, the smaller clusters in the precuneus, middle temporal gyrus, and right insula did not remain significant. These findings are broadly consistent with the sensitivity analysis reported in the original paper, which excluded studies involving individuals at risk of depression and those with remitted depression (see Supplementary Table S3 in the original publication). Collectively, the results support altered dACC activity during autobiographical-memory processing in depression, while indicating that the smaller clusters should be interpreted cautiously (see also the response below).

Moreover, we apologize for insufficient specification of contrasts in the original manuscript. Contrasts were selected from each primary study to capture group differences in the condition serving as the main index of autobiographical memory (typically autobiographical memory recall versus neutral or baseline conditions, as determined by the analytic choices of the original authors; see Table 1 in the original publication for coordinates and tasks). To increase statistical power, multiple contrasts from the same sample were included. Importantly, multiple contrasts from the same sample were handled in line with meta-analytic rules and guidelines, which generally permit the inclusion of multiple contrasts from the same sample when the data are handled accordingly. In line with the developer’s recommendations, multiple contrasts from the same sample were merged into one contrast and counted as one sample (see, e.g., also detailed discussions on handling multiple contrasts in the SDM forum).

Two studies (Keedwell et al., Reference Keedwell, Andrew, Williams, Brammer and Phillips2005; Macdonald et al., Reference MacDonald, Murray, Moutsiana, Fearon, Cooper, Halligan and Johnstone2016) reported center-of-mass rather than peak-voxel coordinates. We agree that this represents an approximation that should have been further explained in the original manuscript. The approach has been previously proposed in the meta-analytic literature (e.g. https://journal.r-project.org/articles/RJ-2014-020/index.html).

Young et al. (Reference Young, Bellgowan, Bodurka and Drevets2013, Reference Young, Bodurka and Drevets2016) share a participant pool, as confirmed by direct email correspondence with K. Young before analysis. As indicated in Table 1 (in the original publication), these publications were treated as a single sample and merged following our pooling procedure (see above). Fourteen studies were included overall, contributing 13 independent participant pools.

Moreover, at-risk and remitted-depression samples were included to enhance statistical power and capture the broader spectrum of autobiographical-memory dysregulation documented in depression; however, the limited number of studies in at-risk and remitted individuals precluded formal subgroup comparisons. Instead – and following discussions with peer reviewers – we conducted a sensitivity analysis restricted to studies of individuals with current major depression (Supplementary Table S3 in the original publication), excluding studies in at-risk and remitted individuals, both of whom were conceptualized as groups with an increased risk of depression. The right dACC (the largest cluster in the primary analysis) remained significant, in agreement with the primary results and the F-statistic sensitivity analysis described above. The smaller clusters in the left precuneus, right middle temporal gyrus, and right insula did not survive. These smaller clusters may reflect more subtle effects rendered nonsignificant under reduced power and should be interpreted with caution pending higher-powered, homogeneous replications in more homogeneous meta-analyses. Of note, and consistent with the main meta-analysis, none of the sensitivity analyses identified alterations in the hippocampus or amygdala. These findings reinforce our central conclusion: the present meta-analysis does not support prevailing models proposing that amygdala and hippocampal abnormalities underlie autobiographical memory deficits in depression. Instead, the results implicate the dACC as a potential neuropathological substrate.

We acknowledge the anterior insula as a core node of the salience network. However, given the multiple functions of the insula and the limited robustness of this cluster in the new sensitivity analyses, an extended discussion of intra-network dynamics would exceed the scope of the present work. We also apologize for a typographical error in the Discussion, where the insula was inadvertently described as ‘hyperactive’; the intended and correct term is ‘hypoactive’, consistent with the Results.

Regions presented in Figure 2b were based directly on the results table from SDM-PSI, which identified the coordinates as located within the corpus callosum. We did not further interpret or discuss the corpus callosum in the main text, given its role as a major white-matter structure and the difficulties in interpreting the meaning of task-related BOLD signal changes in white matter.

We apologize for the unclear reporting of the study count. The correct figures are 14 studies contributing 13 unique participant pools (see also reply above).

We thank Schöpfer et al. once more for their careful engagement with our work. The additional sensitivity analyses reported here strengthen confidence in the central finding of a robust dACC contribution to autobiographical memory dysregulation in depression.