key: cord-0898193-19tfhlsz authors: von Weyhern, Claus Hann; Kaufmann, Ines; Neff, Frauke; Kremer, Marcus title: Neuropathology associated with SARS-CoV-2 infection – Authors' reply date: 2021-01-21 journal: Lancet DOI: 10.1016/s0140-6736(21)00097-0 sha: 7b837be77410b72bfffd43427a0376797f0af583 doc_id: 898193 cord_uid: 19tfhlsz nan Although the global COVID-19 pandemic is a public health emergency and presents us with the need to find out as much as we can about the disease as rapidly as possible, we believe that hastily publishing such findings without adequate peer review should be challenged. Especially in times of description of a new disease, much care should be taken to obtain expert review of the findings and any conclusions need to be substantiated by appropriate evaluations and inclusion of controls. Although this literature draw attention to the important question of CNS involvement of COVID-19, the data presented by von Weyhern and colleagues are likely to cause confusion and possibly misdirect future clinical efforts. We declare no competing interests. The findings of published cases and our own limited experience with COVID-19 disease pathology support that when brain pathology is present, it is apparently often associated with vascular changes, such as thrombi or thromboemboli, rather than primary involvement of the CNS. Although instances with neuropathology consist ent with severe acute respiratory syndrome coronavirus 2 encephalitis have been reported, these changes are generally mild and variable. [2] [3] [4] We declare no competing interests. We welcome the opportunity to respond to the comments about our Correspondence. 1 To increase the availability of data, we reported our findings, which were produced by a team that included an experienced neuropathologist and was subjected to the The Lancet's peer-review process. The three responses are similar as they all challenge our interpretation of the findings presented. Striking is the fact that the three responses disagree among themselves, with each offering yet another interpretation. To analyse our findings, we consulted a standard neuropathology textbook, coauthored by one of our critics. 2 In the textbook, one reads that glial cells indeed can be differentiated from nonglial cells, and that hypoxic neurons can be distinguished from dark neurons based on classical haematoxylin and eosin staining. In the absence of specific morphological alterations, shrinking is often the only indication of necrosis or apoptosis. Aware of the hypoxic origin of similar morphological alterations, we used the general level of hypoxic injury to vulnerable CA1 areas of the hippocampus and Purkinje cells in the cerebellum as a baseline. Because we did not observe hypoxic injury in these areas, we do not attribute the observed morphological alterations to hypoxia. We concede that the term panencephalitis was poorly chosen. In the neuropathological literature, pan-encephalitis refers to a fulminant Sean Gallup/Staff/Getty Images www.thelancet.com Vol 397 January 23, 2021 as they were not presented, and we assumed a counterfactual of people in the education groups "primary or less" and "secondary" moving to "trade, college, or university". Yusuf and colleagues chose less extreme counterfactuals than we have but did not specify them. Nevertheless, the different message conveyed by 6·0% and 57·0% is considerable. Yusuf and colleagues 1 should present PAFs using HRs for each risk factor from models that appropriately account for the temporal relationships between risk factors and cardiovascular disease (eg, based on a directed acyclic graph to select confounders to adjust), 4 and not force PAFs to sum to 100% or less. 5 DL has received support from national and international government and charity funders, Medtronic, and Roche Diagnostics for research that is not related to the content of this Correspondence. All other authors declare no competing interests. necrotic disease, which we clearly did not observe. However, both MacLean Nasrallah and colleagues and a more recent report 3 point out that patients with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) exhibit signs of the observed dis seminated encephalitis as well as aseptic meningitis. Whether this observation is a general feature of SARS-CoV-2 infection should be established through extensive studies of a larger set of patients. It is suggested that we overinterpreted petechial haemorrhage. We agree that petechial haemorrhages are a common finding in autopsies of terminally ill patients, although they are most often localised in the occipital lobe and basal ganglia. In our group of patients, petechial haemorrhaging was located in the frontal and high parietal area and pronounced in the white matter (purpura cerebri-like). Beside the fact that reports of petechial bleeding in the brains of SARS-Cov-2 fatalities are mounting, a fact also alluded to by Nasrallah and colleagues, the massive haemorrhaging observed in our group of patients is certainly not an over-interpretation. [3] [4] [5] [6] [7] Considering the ongoing global pandemic, we are happy that our Corres pondence has generated a lively discussion and can only hope that the various hypotheses discussed in this exchange will be subjected to systematic rigorous analysis in a larger sample. Salim Yusuf and colleagues' 1 findings on the heterogeneity of risk factors associated with cardiovas cular disease in low-income, middle-income, and high-income countries are important. However, hazard ratios (HRs) and population-attributable fractions (PAFs) need to account for causal ordering among risk factors. 2,3 PAFs calculated by use of HRs from regression models that include all risk factors simultaneously violate the principle that causation occurs in sequential steps. The calculated HR is what is left over after adjusting for risk factors that are, at least partly, on the causal pathway leading to cardiovascular disease. For example, Yusuf and colleagues 1 report that only 6·0% (95% CI -4·5 to 16·5) of cardiovascular events in low-income countries are attributable to low education. However, this proportion is the contribution left over after adjusting for diet, smoking, and other proximal factors, through which education also has a causal impact. Using data from the PURE study, 1 we calculated the crude education PAF to be 57·0%. We were unable to use HRs adjusted for confounders (but not mediators) Neuropathology of COVID-19: a spectrum of vascular and acute disseminated encephalomyelitis (ADEM)-like pathology Non-aneurysmal subarachnoid haemorrhage in COVID-19 Microvascular injury and hypoxic damage: emerging neuropathological signatures in COVID-19 COVID-19: a global threat to the nervous system Lifting the mask on neurological manifestations of COVID-19 Early evidence of pronounced brain involvement in fatal COVID-19 outcomes Greenfield's neuropathology We declare no competing interests.