Material

According to the Act on the Investigation of the Cause of Death 1973/459, the Finnish police are required to investigate deaths that entail a possibility of homicide, suicide, accident, medical or surgical adverse event, occupational disease, as well as sudden and unexpected deaths. In most cases, the police order a medico-legal autopsy to be performed by a forensic pathologist in the Forensic Medicine Unit, Finnish Institute for Health and Welfare. A medico-legal autopsy is performed in approximately 15% of all deaths in Finland [9, 10]. Approximately 1 in 40 medico-legal autopsies involves a full neuropathological examination of a formalin-fixed brain performed by a board-certified neuropathologist [11].

The Helsinki office of the Forensic Medicine Unit covers Southern Finland and the capital metropolitan region, totaling approximately 3000 medico-legal autopsies annually. An electronic information system was adopted in 2016 as a comprehensive storage for medico-legal cause-of-death investigation documents. For the purposes of this study, a member of the research group accessed the documents of the cases that fulfilled the following inclusion criteria:

(a)

A medico-legal autopsy performed at the Helsinki office in 2016—2023.

(b)

The general autopsy supplemented with a full neuropathological examination performed by a consultant neuropathologist.

(c)

β-APP staining performed in the neuropathological examination to diagnose AI.

The data collection covered all cases that fulfilled the inclusion criteria; there were no exclusions. The study was performed in accordance with the Declaration of Helsinki and national legislation on medical research. As the study was retrospective and based on register-based data only, ethical approval was not required. A research permit was obtained from the institutional data permit authority of the Finnish Institute for Health and Welfare (THL/1802/6.02.00/2023).

Axonal injury status in β-APP stain

The consultant neuropathologist’s laboratory is part of a large, public, tertiary-level pathology department. A comprehensive neuropathological examination of the formaldehyde-fixed brain (minimum of two weeks) was performed by a board-certified neuropathologist. The examination included β-APP-stained samples (Anti-APP A4 Antibody, MAB348) of the corpus callosum, internal capsule, and brainstem, as a minimum standard. The aetiology of AI was estimated on the basis of the β-APP staining pattern. In brief, scattered β-APP-positive axons involving single white matter tracts are considered characteristic of TAI, whereas linear and geographical patterns of positive axons are regarded to represent VAI [2]. Figure 1 illustrates the typical staining patterns consistent with TAI and VAI.

Fig. 1

Sample micrographs demonstrating a typical staining pattern in traumatic axonal injury (A) and vascular axonal injury (B) in β-amyloid precursor protein stain. Both samples involve the pons, 40x magnification. Scale bars equal 0.1 mm

The neuropathologist’s report was used to classify cases into the following AI categories:

1)

Staining pattern suggestive of TAI (“TAI” group).

2)

Staining pattern suggestive of VAI (“VAI” group).

3)

No staining positivity suggestive of AI (“No AI” group).

Cases were classified as TAI or VAI regardless of whether the finding was focal, multifocal, or diffuse. To make categorizations mutually exclusive, each case was classified into one category only. Importantly, it was taken into consideration that TAI and VAI were expected to co-occur quite often in traumatic cases [4, 8]. If a case was reported to have a TAI pattern in any of its samples, it was classified into the TAI group even if secondary VAI was also reported. As such, the TAI group comprised cases with “pure TAI” and co-occurring TAI + VAI, whereas the VAI group only comprised cases with “pure VAI”. This decision was based on the fact that the main goal of a forensic pathologist is to distinguish traumatic cases (TAI with or without VAI) from non-traumatic ones (pure VAI).

For TAI and VAI cases, the location of AI was recorded (brainstem/internal capsule/corpus callosum/other/unspecified; yes/no for each). In TAI cases, also the diagnosis and grading of diffuse TAI (dTAI) was recorded according to Adams et al. [7].

Background characteristics

The medico-legal autopsy order of the police was used to record the sex (male/female) and age of the decedent (years). Autopsy order and medical records were used to extract medical history regarding the central nervous system (CNS) (underlying diseases, surgical procedures, and remote injuries). Underlying conditions of the CNS were categorized in a data-driven manner (epilepsy or history of seizures/dementia or neurodegenerative disease/stroke or transient ischaemic attack/neoplasm of the CNS/hydrocephalus/other).

The autopsy order and medical records were also queried for a documented recent injury or asphyxic event (yes/no for each). In case a recent injury was documented, its general circumstance (fall/traffic accident/assault/other) and postinjury survival (< 30 min/30 min–23 h/≥ 24 h/unknown) were recorded. Postinjury survival was estimated as accurately as possible on the basis of the aforementioned documents.

The manner and primary cause of death were recorded from the death certificate issued by a forensic pathologist. The manner of death was recorded as disease/accident/homicide/other, and the primary causes of death were categorized in a data-driven manner (head or neck injury/stroke/cardiovascular disease/intoxication/asphyxiation/sudden infant death syndrome/other). Postmortem interval (days) was calculated by taking the difference between the known or estimated time of death and autopsy date.

Autopsy and neuropathology findings

The general medico-legal autopsy report was used to record the following variables: fresh brain weight (g) of the decedent, presence of external injury to the head (laceration, stab wound, bruise, traumatic swelling), scalp haemorrhage, and fracture of the cranium, facial bones, or cervical spine (yes/no for each).

The presence of acute traumatic intracranial haemorrhages (epidural, subdural, subarachnoid, intracerebral/ventricular) were recorded from both the general autopsy and neuropathology reports (yes/no for each). The general autopsy report was used as the main source for epidural and subdural haemorrhage, and the neuropathology report for subarachnoid and intracerebral/ventricular haemorrhage. Contusions were documented from neuropathology reports.

The neuropathology report was also used to record the following variables (yes/no for each): brain oedema (macroscopic or microscopic), herniation, hypoxic-ischaemic neuronal injury (pyknosis and/or eosinophilia in haematoxylin-eosin stain; mostly mild and terminal), cerebral atrophy (cortical or central), vermal atrophy, neuropathological changes warranting a neurodegenerative disease diagnosis, atherosclerosis in the circle of Willis, arteriolosclerosis, cerebral amyloid angiopathy, and brain infarction (micro, lacunar, or major). All findings were recorded as reported by the neuropathologist, regardless of severity.

Statistical analysis

The statistical analysis was performed in SPSS version 27 (IBM, Armonk, NY, USA). For continuous variables, medians were calculated and reported with interquartile ranges (IQRs). For categorical variables, prevalences were calculated for the three AI groups (TAI, VAI, no AI) and presented as percentages (%) with frequencies (n).

Comparisons between the three AI groups were based on categorical variables. The statistical differences in prevalences between the three AI groups were analyzed using the Chi square test and Fisher-Freeman-Halton Exact Test, as applicable. In order to offer the reader a better sense of effect sizes, pairwise head-to-head comparisons were performed on the basis of prevalence ratios (e.g., prevalence among TAI cases divided by prevalence among VAI cases). Statistical and clinical relevance was defined as follows: a statistically significant difference between groups (P < 0.05) combined with a prevalence ratio of > 2 or < 0.5, which indicates more than a two-fold difference in prevalence between groups. Prevalence ratios were illustrated on a logarithmic scale using graphs created in Excel version 2005 (Microsoft, Redmond, WA, USA).