Introduction

In the field of forensic medicine, postmortem interval (PMI) estimation is crucial. The postmortem changes in the body, including body temperature, rigor mortis, lividity, and corneal opacity, are useful to estimate PMI until approximately 72 h postmortem1. However, body temperature, rigor mortis, and lividity are strongly influenced by external factors. Corneal opacity determination with the naked eye is subjective. Therefore, objective evaluation criteria that are not significantly influenced by the external environment are worthy of further exploration.

Because the eyeball is protected by the orbital cavity and the closed eyelid, postmortem changes in the eyeball are not significantly influenced by the external environment. In recent years, an increasing number of studies have been conducted on the correlation between PMI and postmortem eyeball changes, particularly in corneal opacity2, corneal thickness3, iris and pupil4, lens5, retina6, intraocular pressure7, eye temperature8,9, and vitreous electrolyte concentrations6. However, these examinations require invasive procedures or expensive instruments. To address this problem, the development of more convenient and noninvasive methods is urgently needed.

The corneal epithelial cells are divided into three layers from the anterior to the posterior, namely, superficial, intermediate, and basal cells. A previous study on the maximum time limit postmortem for harvesting donor cornea reported that the epithelial cells shed layer-by-layer with the prolongation of PMI10,11. Observation of postmortem changes in the epithelial layers may be useful for the estimation of PMI in forensic medicine. However, in forensic practice, as the cornea of a cadaver cannot be collected before the autopsy, histological examination of corneal tissue sections is not feasible.

In ophthalmology, impression cytology is mainly employed in the screening of dry eye, keratoconjunctivitis and in the noninvasive detection of ocular surface tumors12. This method was developed by Egbert in 197713. It involves harvesting epithelial cells by applying a membrane filter to the cornea or conjunctiva for a few seconds under eye-point anesthesia. Impression cytology is simple and causes minimal damage to the cornea and conjunctiva. The corneal epithelium serves as a barrier, protecting the eye from the external environment, and this barrier function is maintained by tight junctions, which are cell–cell adhesive structures. The tight junctions of the corneal epithelium are composed of tight junction proteins, including ZO-1 and occludin. In the immunocytochemistry of intact human cornea, ZO-1 was restricted to the most superficial cells and was not observed in the intermediate or basal epithelial cells14. In the immunohistochemistry of the transverse sections of the human cornea, ZO-1 was stained at most apical superficial cells, as well as the border between the basal and intermediate cells15. Meanwhile, occludin was apparent in the superficial and intermediate cell layers of the epithelium in the transverse sections of the rat cornea16. Ban et al. demonstrated that occludin was concentrated at the cell–cell borders only in the superficial layer and that it appeared to be expressed in the superficial and intermediate cell layers in the human cornea15. Immunofluorescence staining with anti-ZO-1 and anti-occludin antibodies may be able to identify the epithelial cells harvested via impression cytology. This study aimed to determine whether corneal epithelial cells collected from the eyes of a cadaver via impression cytology are useful for the observation of postmortem changes in the corneal epithelium.

Results

Immunofluorescence staining of corneal epithelial cells collected from cadavers with different PMIs

The representative images of double-staining with anti-ZO-1 and anti-occludin antibodies are presented in Fig. 1a. Most of the cells in group 1 exhibited a pentagonal or hexagonal shape and were close to each other. Either ZO-1 or occludin was stained on the edge of each cell, and the stain was concentrated on the regions of contact between the cells (Fig. 1b). The cells shrank and scattered with the prolongation of the PMI.

Fig. 1
figure 1

Panel (a) shows representative images of the immunofluorescence staining of corneal epithelial cells collected from cadavers with different postmortem intervals. NC, negative control; scale bar = 50 µm. Panel (b) shows magnified images of the cells in group 2.

Full size image

ZO-1 (+)/occludin (+) cells accounted for 98.8% (median) of the total number of cells in group 1 (Fig. 2a). This value did not statistically change in group 2 but gradually decreased with the prolongation of the PMI to 78.4% in group 6. Spearman’s rank correlation coefficient was − 0.69, indicating a negative correlation. Contrarily, the percentage of ZO-1 (−)/occludin (−) cells was only 0.39% in group 1, which increased with the prolongation of the PMI in group 4 and reached 19.3% in group 6 (Fig. 2c). A strong positive correlation was observed between the percentage of ZO-1 (−)/occludin (−) cells and PMI (Spearman’s rank correlation coefficient = 0.70). Meanwhile, the percentage of ZO-1 (−)/occludin (+) cells initially increased in group 3 but subsequently decreased (Fig. 2b). Spearman’s rank correlation coefficient was 0.20.

Fig. 2
figure 2

Relationship between the postmortem interval and the percentages of ZO-1 (+)/occludin (+) cells (a), ZO-1 (−)/occludin (+) cells (b), and ZO-1 (−)/occludin (−) cells (c). Blue and red dots denote the male and female cases, respectively. Box plots show the minimum, first quartile, median, third quartile, and maximum values. rs, Spearman’s rank correlation coefficient. Steel’s multiple comparisons test was employed to compare the 6–12 h postmortem group with the other groups; the analytical results are presented at the top of the graph. n.s., not significant; **P < 0.01; *P < 0.05.

Full size image

The influence of other factors on the ZO-1/occludin-positive rate of postmortem corneal epithelial cells

To evaluate the simultaneous effect of PMI and another independent variable, including age, gender, and wet condition, on postmortem ZO-1/occludin-positive cell rate, two-way ANOVA analyses were conducted on the percentages of ZO-1 (+)/occludin (+) and ZO-1 (−)/occludin (−) cells, which statistically correlated with PMI. First, 114 cases, including those in their 50 s (50–59 years old, n = 23), 60 s (60–69 years old, n = 22), 70 s (70–79 years old, n = 36), and 80 s (80–89 years old, n = 33) were analyzed for the interaction between PMI and age on each cell type (Table 1). The effect of PMI was statistically significant, but that of age was not in both cell types. No significant interaction was observed between PMI and age.

Table 1 P-value in two-way ANOVA between the postmortem interval (PMI) and age on the ZO-1/occludin-positive cell rate.
Full size table

Second, 134 cases, including 89 men and 45 women, were analyzed for the main effect of gender (Table 2). No significant effect of gender was observed in both cell types. However, there was a significant interaction between PMI and gender in ZO-1 (−)/occludin (−) cells but not in ZO-1 (+)/occludin (+) cells.

Table 2 P-value in two-way ANOVA between postmortem interval (PMI) and gender on the ZO-1/occludin-positive cell rate.
Full size table

Wet and dry conditions may influence postmortem changes in the corneal epithelium. Cases until 48 h postmortem were divided into three groups based on the situation in which the cadaver was found: with the face submerged in the water in the bathroom (n = 12), with the face not submerged in the water in the bathroom (n = 8), and somewhere other than the bathroom or the waterside (n = 61). No significant effect of wet conditions and no significant interaction between PMI and wet conditions were observed in both cell types (Table 3).

Table 3 P-value in two-way ANOVA between postmortem interval (PMI) and wet condition on the ZO-1/occludin-positive cell rate.
Full size table

Discussion

The majority of the cells collected from cadavers 6–12 h postmortem via impression cytology were identified as superficial cells based on the shapes and immunofluorescence staining patterns. Examination of 51 human donor corneas by scanning electron microscopy or light microscopy revealed intact epithelium within 48 h postmortem when the body was placed in a cold storage room at 4 °C within a few hours after the death11. After 48 h, superficial and deep epithelial layer separation, as well as epithelial cell loss, increased. Consistent with these findings, the present study demonstrated that the percentage of ZO-1 (+)/occludin (+) cells decreased at 25 h postmortem and that of ZO-1 (−)/occludin (−) cells increased at 49 h postmortem.

Meanwhile, the percentages of ZO-1 (−)/occludin (+) cells initially increased at 25–48 h postmortem and then subsequently decreased. In forensic pathology, a decline in immunohistochemistry reaction is frequently observed in tissue samples obtained from decomposed bodies. In a tissue microarray with 120 forensic autopsy samples, including liver, lung, and brain tissues, samples from cadavers collected 1–3 days postmortem exhibited positive staining with all antibodies, whereas samples with a longer PMI showed decreased staining rates17. In immunocytochemical staining of human, guinea pig, and rat colon and urinary bladder samples left for 48 h at 20 °C until fixation, the quantity of positive staining substantially decreased owing to the autolysis of the tissue18. Immunohistochemistry in human surgical tissues revealed that the variation in staining intensity was mainly affected by the time interval prior to tissue fixation, and the staining intensity loss became evident at 48 h19. The number of ZO-1/occludin negative cells was likely to be partially affected by the decline in stainability with the antibodies.

Regarding age-related changes in the corneal epithelium, there have been several reports on corneal epithelial thickness (CET)20,21, corneal epithelial cell density, and corneal epithelial cell area22. In the present study, among the 50–89-year-old groups, the changes in the corneal epithelium were mainly affected by PMI but not by age. These inconsistent findings could be attributed to different measurements. Furthermore, the age range in the present study was narrower than those in the other studies and did not include cases below 50 years old. Although the prevalence of dry eye increases with age23, it could not be determined whether every cadaver has some eye disease.

No statistically significant difference was observed between men and women in the CET24. Meanwhile, other studies reported that men had thicker corneas than women21,25. In the present study, gender had no effect on the postmortem changes in the corneal epithelium, whereas an association was observed between gender and PMI in ZO-1 (−)/occludin (−) cells. This could be attributed to the significant difference in the number of cases between men and women as well as the small number of cases in general.

In mice exposed to a low-humidity environment, corneal barrier dysfunction was induced, which resulted in corneal epithelial shedding26. However, it is unclear whether the effect of low humidity on corneal barrier function is extended to postmortem eyes. In the present study, the postmortem change in the corneal epithelium seemed to be unrelated to the situation in which the cadaver was found: in the bathroom or somewhere other than the bathroom or the waterside. However, as the present study was conducted during the high-humidity season in Kyoto, no significant difference was observed. Furthermore, although the eyelids of the cadaver were kept closed at least while being transported to our department, we could not obtain information on whether the eyelids were opened when the body was found.

To our knowledge, impression cytology has never been employed in forensic practices. It is suitable for forensic examination as it is noninvasive and fast method and does not affect the subsequent examination or autopsy. The immunofluorescence technique of impression cytology was successfully adapted by Pflugfelder et al. in 199027. In the present study, the membrane filter was dissolved with acetone before immunofluorescence staining according to Krenzer and Freddo’s protocol28 to eliminate background fluorescence and obtain better fluorescence images.

This study has several limitations that need to be acknowledged. The first is the possible contamination of conjunctival epithelial cells and blood cells in the impression cytology. Either ZO-1 or occludin was positively stained in conjunctival apical superficial cells and corneal epithelial cells29. Thus, we excluded cases with blood on the cornea and in the conjunctival sac from the analysis. When we performed impression cytology, we opened the eyelid as gently as we could and placed the membrane filter on the center of the cornea. In addition, washing the bulbus with saline or other aqueous sterile solution before impression cytology may have prevented contamination of the sample with other cell types. The second limitation was that we could not fully obtain information on the situation when the body was found. In particular, ambient temperature and whether the eyelids are open or closed can have a big impact on corneal epithelial condition. The subject’s history of eye disease was also unknown. In an injured corneal epithelium, ZO-1 expression was induced in the intermediate30,31 and basal31 cells to reform tight junctions during recovery. Finally, in this method, 3 h may be required to dissolve the membrane for immunofluorescence staining. Forensic applications may also require methods that do not necessitate the dissolution of the membrane for faster processing32,33,34.

In conclusion, the present study demonstrated that it is possible to observe PMI-dependent changes in the corneal epithelium through immunofluorescence staining of tight junction proteins on corneal epithelial cells collected from the eyes of cadavers via impression cytology. A simple and noninvasive impression cytology may be applicable for diagnosis in forensic practice.

Methods

The Ethical approval

This study was approved by the Institutional Review Board of the Kyoto Prefectural University of Medicine (ERB-C-3107). All experiments were performed in accordance with the Declaration of Helsinki. Sources of cadavers were medico-legal autopsies commissioned by the Kyoto Prefectural Police. Moreover, this study was conducted using autopsy records from the past. However, we could not obtain informed consent from the bereaved family for the use of these records. Therefore, we conducted this study in accordance with the Ethical Guidelines for Medical Research Involving Human Subjects, Chapter 5, Part 12-1, (b), (ii) (enacted by the Ministry of Health, Labor, and Welfare in Japan). We posted information about our research on our website (https://square.umin.ac.jp/kpum-hoi/research.html), allowing for the exclusion of individuals whose relatives did not agree to the study participation. The Institutional Review Board of the Kyoto Prefectural University of Medicine approved this method and waived the need for informed consent.

Study subjects

The subjects for the impression cytology were selected among cadavers taken to our department for postmortem computed tomography (CT) and/or autopsy from March 15 to August 31, 2024, based on the inclusion/exclusion criteria. The cadaver was transported to our department at room temperature with closed eyelids. Impression cytology was performed within approximately 15 min after the arrival of the cadaver. The inclusion criteria were PMI < 2 weeks and age > 18 years. The exclusion criteria were a high degree of decomposition, burning, eyeball atrophy, corneal dryness, eyeball injury, palpebral fissure stenosis, conjunctival healing, and blood or mucus on the cornea and in the conjunctival sac. The interval from the estimated time of death to the start time of CT examination or autopsy was defined as PMI. Impression cytology was performed in 134 cases that met the inclusion and exclusion criteria. The number of male and female cases were 89 and 45, respectively. The age range was 21–97 years, with a median age of 71 years. The PMI ranged from 6 h to 14 days. These cases were divided into six groups based on their PMIs (Table 4).

Table 4 Number of cases in each group divided according to the postmortem interval.
Full size table

Impression cytology

Cellulose membrane filter (GSWP03700, Merck Millipore, Tokyo, Japan) was cut into a right trapezoid (2-mm long at the top, 4-mm long at the bottom, and 5-mm high). The membrane filter was placed on the central surface of the cornea using forceps and lightly pressed for 5 s using a small stick13. Subsequently, the picked membrane filter was sprayed with Cytosetter (J20000, Matsunami glass, Osaka, Japan) for fixation and was put the cell-attachment side down on a poly L-lysine-coated glass slide (Muto Pure Chemical, Tokyo, Japan). After drying the membrane filter, a drop of acetone was applied to facilitate the adhesion of cells to the glass slide, and the membrane filter was dried again to ensure complete adhesion. Then, to dissolve the membrane filter, the slide was soaked in acetone solution with agitation at 30 rpm for 1 h. After washing the slide with tap water for 5 min, the membrane filter residue was digested with 10-U/mL cellulase (Nacalai Tesque, Kyoto, Japan) solution (0.1 M acetate buffer pH 5) for 2 h at 37 °C28. After washing with tap water followed by phosphate-buffered saline (PBS), the specimen was blocked with 5% donkey serum (D9663, Sigma-Aldrich, Tokyo, Japan) in PBS for 30 min and then incubated with anti-ZO-1 (1:1000, 21773-1-AP, Proteintech, Tokyo, Japan) and anti-occludin (1:2000, 66378-1-lg, Proteintech) antibodies for 1 h at room temperature. Next, after washing with 1% BSA with PBS three times, the specimen was incubated with ZO-1 (1:500, Goat Anti-Rabbit IgG H&L Alexa Fluor® 488, ab150081, Abcam, Tokyo, Japan) and occludin (1:1000, Goat Anti-Mouse IgG H&L Alexa Fluor® 568, ab175701, Abcam) antibodies for 1 h at room temperature. Afterward, the specimen was mounted with medium with DAPI (VECTASHIELD #H-1800, Vector Laboratories, Newark, CA). A total of 12 fluorescence images were obtained from each slide using a fluorescence microscope (BZ-X710 All-in-One Fluorescence Microscope, Keyence, Osaka, Japan) at a magnification of 40×. The specimen not stained with the primary antibody was used as the negative control. The DAPI-stained cells were classified into ZO-1 (+)/occludin (+), ZO-1 (−)/occludin (+), and ZO-1 (−)/occludin (−). The number of cells was counted using an imaging software (cellSens 1.16, Evident, Tokyo, Japan, https://evidentscientific.com/en/products/software/cellsens).

Statistical analysis

Quantitative data were expressed as median (interquartile range [IQR]) and analyzed using EZR35. Spearman’s rank correlation coefficient was used to determine the correlation between PMI and the number of immunostained cells. Steel’s multiple comparisons test was employed to compare the 6–12 h postmortem group with the other groups. The influence of the other factors on the correlation was investigated via two-way analysis of variance (ANOVA) within the different PMI groups. P-values of 0.05 or less were considered to be statistically significant.

Conference presentation

This study was presented at a regional congress of Japanese Society of Legal Medicine in Nara, Japan (11/9/2024). The short abstract will be published on the website of Japanese Society of Legal Medicine.