Feasibility of using tissue autolysis to estimate the postmortem interval in horses

Abstract

Estimation of the postmortem interval (PMI) is a poorly studied field in veterinary pathology. The development of field-applicable methods is needed given that animal cruelty investigations are increasing continually. We evaluated various histologic criteria in equine brain, liver, and muscle tissue to aid the estimation of PMI in horses, which is central to forensic investigations of suspicious death. After death, autolysis proceeds predictably, depending on environmental conditions. Currently, no field-applied methods exist that accurately estimate the PMI using histology in animals or humans through quantification of autolysis. Brain, liver, and skeletal muscle from 12 freshly euthanized horses were held at 22°C and 8°C for 72 h. Tissues were sampled at T0h, T1h, T2h, T4h, T6h, T12h, T24h, T36h, T48h, T60h, and T72h. For each tissue, we quantified 5 to 7 criteria associated with autolysis, based on the percentage of microscopic field involved. Each criterion was modeled, with temperature and time as independent variables. Changes were most predictable in liver and muscle over the first 72 h postmortem. The criteria for autolysis that were present most extensively at both temperatures were hepatocyte individualization and the separation of bile duct epithelium from the basement membrane. The changes that were present next most extensively were disruption of myofiber continuity, hypereosinophilia, and loss of striation. Brain changes were highly variable. The high statistical correlation between the parameter “autolysis” and the variables “time/temperature”, indicates that autolysis is progressive and predictable. Further investigation of these criteria is needed to establish histologic algorithms for PMI.

After death, the blood supply to tissue compartments is compromised, causing acute hypoxic cell injuries that manifest as acute cellular swelling, hydropic and vacuolar degeneration, reduction in adenosine triphosphate (ATP), membrane damage, increased intracytoplasmic calcium, and formation of reactive oxygen species, all leading to cell death and initiation of autolysis (self-digestion).^8,9 After death, this process of cell death is postmortem autolysis, the earliest step of decomposition, which is a process of denaturation of intracellular proteins and nucleic acid, protein coagulation, and enzymatic digestion of nuclei and cytoplasm (Table 1). Enzymes are released postmortem from the swollen lysosomes of dying cells by diffusion through the lysosomal membrane.^9,11 More specifically, the enzymes driving autolysis are mostly phosphorus-rich enzymes including alkaline and acid phosphatases, ATP, 5′-nucleotidases, and glucose-6-phosphatase.^11,18

Table 1.

Comparison of the features of autolysis, necrosis, and apoptosis.

	Autolysis	Necrosis	Apoptosis
Occurs in living tissue	No	Yes	Yes
Occurs only postmortem	Yes	No	No
Programmed cell death	No	No	Yes
Occurs in healthy tissue	Yes	No	Yes
Physiologic process	NA	No	Yes
Lethal injury necessary	No	Yes	No
Accompanied by inflammatory reaction	No	Yes	No
Enzymatic digestion	Yes: enzymes released from the dying cell itself	Yes: enzymes released by surrounding leukocytes	No
Harmful effects on surrounding tissue	Yes	Yes	No
Caused by infectious or toxic agents	No	Yes	Yes

NA = not applicable.

Histologically, the morphologic changes of cell autolysis are similar, if not identical, to those of cells undergoing necrosis, which is focal-to-extensive, non-programmed cell death that occurs after irreversible exogenous injury to a cell (Table 1). Necrosis is also a process of progressive enzymatic cell death, but within living tissue, and denaturation of intracellular proteins is caused mostly by lysosomal enzymes released from immigrant leukocytes, which are major actors in any inflammatory process.⁸ Necrosis can be initiated by a lack of blood supply as described above, as well as by infectious agents (bacteria, viruses, parasites, and protozoans) or toxins, which all, more often than not, cause inflammation.^{6,8,9,13,19,20}

A third type of cell death is apoptosis, a programmed process that occurs most often in healthy tissues and is considered a physiologic event in which unwanted cells are eliminated but can occasionally be pathologic after some forms of cell injury (Table 1).⁸ Detailed descriptions of autolysis and necrosis are available in the literature.^8,10

Using light microscopy, necrosis and autolysis have similar morphologic changes, except that autolysis occurs without inflammatory cell infiltrates. Autolysis is also seen in the entire organ or body, occurring usually at the same rate, and occurs in tissues that have not been fixed immediately after the patient’s death. Cells or tissues placed immediately in fixative are dead but not necrotic, if fixation is prompt and adequate.^8,9 Focal cell death observed adjacent to normal-appearing tissue in the same section suggests that this cell death is necrosis and not postmortem autolysis.⁹ Unfortunately, autolysis can be present in an irregular and patchy distribution. Furthermore, red blood cells often appear hemolyzed in autolyzed tissues, suggesting a postmortem event.^8,9 Necrotic tissue present in tissue that has not been fixed rapidly after death will also undergo postmortem autolysis.⁹ Different cell types undergo postmortem autolysis at different rates because certain cells are more resistant to hypoxia than others.^1–3,5,8 The intestinal mucosa, gall bladder, pancreatic parenchyma, and adrenal medulla autolyze first, then neurons are the next most susceptible, with connective tissue the slowest to undergo degradation.^1–3,5,8,10 Given that much of this activity is enzymatically driven, the rate of autolysis as well as its evolution to bacterial putrefaction, is slowed by refrigeration.

Histologic changes of autolysis during the early postmortem interval (PMI) result from alteration of cell membrane permeability brought about by ischemia and cessation of metabolic activity.^8,9 Autolysis can resemble necrosis. Necrosis is commonly accompanied by other pathologic changes, including a visible inflammatory reaction and/or formation of coagulative material, liquefaction, mummification, caseous debris, mineral deposits, and fibrin.^6,8,13,19,20 Inflammation does not occur in autolysis. Autolytic changes are very familiar to histopathologists, and these changes can interfere with accurate histologic evaluation. Criteria are limited for quantifying autolysis and hence to estimate the time of death.^{7,11,12,15–17} Postmortem changes in the liver have been described in guinea pigs.¹⁶ Neuronal changes in humans are described in a handful of studies.^11,12,15 Autolytic postmortem changes have been described in postmortem traumatized skeletal muscle in dogs¹⁷ and humans.⁷ Autolytic changes are not currently used by medical examiners to declare time since death in humans (personal communications). Importantly, to our knowledge, very limited work describes quantitatively the progression of the intensity and distribution of autolytic changes in any species.

Historically, autolysis has been associated with time since death,⁸ given that it progresses continuously over time; however, histologic evaluation has not been used to systematically quantify histologic changes over exact time intervals, especially in species that are common subjects of forensic investigations. One such species commonly investigated is the horse because of the value placed on sport, performance, and breeding. Although forensic analysis of suspicious equine deaths is a common event as a result of insurance claims, malicious treatment, theft, and overall value of these animals, horses are rarely the subject of forensic research. However, time of death, an essential component of each investigation, remains a challenge to estimate if the death was not witnessed (e.g., euthanasia).

We investigated the hypothesis that specific histopathologic changes associated with autolysis would occur in a time- and temperature-dependent manner in equine tissues. Our major aim was to perform a systematic analysis of changes associated with autolysis of equine tissues. Tissues from 12 horses were utilized to achieve a power of ≥0.80, and liver and muscle were chosen based on publications demonstrating that liver was least stable whereas muscle was most stable over time.⁸ Brain was included because there was little information regarding degradation of brain during PMI and, given the large size of the horse brain, we wondered if autolysis could be more predictable in this species.

More often than not, the literature describes the use of a limited number of or only one animal, limited sample times used for analysis, and variable length of analysis as a result of insufficient tissues being available.^14,17 We designed our study to provide a more comprehensive analysis of autolysis in equine tissues. We quantified histomorphologic criteria of autolysis in the equine brain, liver, and skeletal muscle during the first 72 h (3 d) of the PMI. Criteria were based on changes described previously.^8,11,18

Materials and methods

Study design

Twelve healthy horses (Table 2) without clinical signs, in adequate body condition (adequate subcutaneous and intra-abdominal adipose tissue), surrendered by owners because of loss of use, were euthanized using sedation with a combination of acepromazine HCl (2–4 mg/45 kg) and xylazine (0.5 mL/45 kg) intravenously, followed by an overdose of pentobarbital and phenytoin (10 mL/45 kg, Beuthanasia-D; Merck Animal Health). Each animal was declared dead once ocular and palpebral reflexes ceased. All animal work was performed with approval of the University of Florida Institutional Animal Care and Use Committee (201004309).

Table 2.

Signalment of horses used in our study of autolysis.

Horse	Sex	Age (y)	Weight (kg)
1	M	21	U
2	F	12	U
3	M	U	496
4	M	U	473
5	M	19	570
6	F	15	510
7	M	22	485
8	U	U	464
9	U	15	510
10	M	4	340
11	M	4	470
12	F	11	590

F = female; M = male; U = unknown.

After exsanguination, a large sample (10 × 20 × 20 cm) of semitendinosus muscle, another large sample (20 × 20 × 5 cm) of liver, and the entire brain were removed from each carcass. Each tissue was bisected, and one-half was maintained at ~22°C, the measured room temperature of the autopsy facility; the other half was maintained at 8°C, the measured temperature of the autopsy walk-in cooler. The organs were kept in large, sterilized containers and covered with aluminum foil. Tissues were sampled for histopathology at “time 0 hour” (T0h), T1h, T2h, T4h, T6h, T12h, T24h, T36h, T48h, T60h, and T72h. The time between death and T0h was 40 min for all horses. Brain samples were taken only from the cerebrum, starting at the rostral lobe, throughout the entire cerebrum to the caudal lobe, which provided the T72h samples. A 0.5-cm slice of each tissue was placed in a tissue cassette and fixed for 24 h in 10% neutral-buffered formalin, followed by immersion in 75% ethanol. Formalin-fixed tissues were processed routinely, and 4-µm sections stained with hematoxylin and eosin.

Histopathology and tissue criteria

The slides of each tissue held at each temperature were blinded and read twice on a light microscope (DMLS Dual VI; Leica) by one of the authors (N. Wenzlow), a trained veterinary pathologist. All slides from fresh tissues at T0h were evaluated for the absence of antemortem tissue necrosis or other lesions that might interfere with the interpretation of autolysis in tissue samples from later times (Suppl. Fig. 1).

Criteria of autolysis (Suppl. Table 1) were selected based on literature review.^{7,11,12,14–18} Five representative fields per slide were evaluated at 200× magnification for each criterion of autolysis. In each field, the presence or absence of each criterion was recorded and graded in the following 6 categories: grade 0 if a criterion was absent, grade 1 if a criterion was present in <20% of the field, grade 2 if a criterion was present in 21–40% of the field, grade 3 if a criterion was present in 41–60% of the field, grade 4 if a criterion was present in 61–80% of the field, and grade 5 if a criterion was present in 81–100% of the field.

Statistical analysis

The grading of all 5 fields was averaged for each slide. Then, the grades from all horses were averaged for each criterion for each time. Data were grouped by tissue and temperature and treated as non-repeated measurements. For all analyses of the histologic criteria, the data were graphed to examine trends over time for each criterion. If a pattern was not discerned, correlation between time and outcome was determined using the Spearman rank correlation coefficient. When a pattern was apparent, linear, parabolic, square root, or exponential equations were derived depending on the shape of the graph. The best-fitting line was modeled, and a p-value calculated using R v.3.2.4 (https://www.r-project.org/).

Results

Liver

In samples held at 22°C, we noted linear increases in hepatic plates separating from one another, leaving clear spaces between the rows of hepatocytes (hepatocyte individualization, p ≤ 0.0001; Fig. 1), bile duct epithelial cell separation from the basement membrane (p = 0.0001; Fig. 2), and bile duct epithelial pyknosis (p = 0.030; Fig. 3).

Figure 1.

Hepatocyte individualization in the livers of 12 horses held at 22°C over 72 h postmortem. A. Each point is the average field involvement of the criterion in 5 fields at 200× magnification at each time (0 = absent, 1 = 1–20%, 2 = 21–40%, 3 = 41–60%, 4 = 61–80%), p < 0.0001. B. Hepatocyte individualization (arrows) in a horse liver at 60 h. H&E. 200×. Inset: hepatic plates separated from each other. 400×.

Figure 2.

Bile duct epithelium separation from its basement membrane in the livers of 12 horses held at 22°C over 72 h postmortem. A. Each point is the average field involvement of the criterion in 5 fields at 200× magnification at each time (0 = absent, 1 = 1–20%, 2 = 21–40%), p = 0.0001. B. Bile duct epithelium separation from its basement membrane (arrows) in a horse liver at 60 h. H&E. 400×.

Figure 3.

Bile duct epithelium pyknosis in the livers of 12 horses held at 22°C over 72 h postmortem. A. Each point is the average field involvement of the criterion in 5 fields at 200× magnification at each time (0 = absent, 1 = 1–20%), p = 0.0001. B. Bile duct epithelium pyknosis (arrow) in a horse liver at 72 h. H&E. 200×.

In samples held at 8°C, we noted increases over time in variably sized, irregular empty vacuoles in hepatocellular cytoplasm (hepatocyte vacuolation, p = 0.02; Suppl. Fig. 2) and bile duct epithelium separation from its basement membrane (p = 0.0003; Suppl. Fig. 3). The correlation for hepatocyte vacuolation was best fit as the square root over time; the bile duct epithelium separation from its basement membrane increased linearly over time. For all remaining criteria, there was no significant trend over time.

In the liver of 4 horses (22, 21, 19, and 15 y old), there was mild inflammation independent of autolysis and consisting of minimal-to-mild, mostly portal, and occasionally random, mixed-cell inflammatory infiltrates (Suppl. Fig. 4). This was considered an incidental, age-related finding, and did not interfere with our evaluation of autolysis.

Skeletal muscle

For skeletal muscle maintained at 22°C, we noted linear increases over time of shortened myofibers that left empty spaces, referred to as interruption of fiber continuity (p = 0.013; Fig. 4A, 4C), increased myofiber eosinophilia (p = 0.003, Fig. 4B, 4C), loss of striation (p = 0.003; Fig. 5A, 5C), and a floccular and granular aspect of myofibrils referred to as sarcoplasmic fragmentation (p = 0.046; Fig. 5B,5C).

Figure 4.

Interruption of myofiber continuity and myofiber eosinophilia in semitendinosus skeletal muscle of 12 horses held at 22°C over 72 h postmortem. A. Each point is the average field involvement of the criterion in 5 fields at 200× magnification at each time (0 = absent, 1 = 1–20%, 2 = 21–40%), p = 0.013. B. Each point is the average field involvement of the criterion in 5 fields at 200× magnification at each time (0 = absent, 1 = 1–20%), p = 0.013. C. Interruption of myofiber continuity (arrows) and skeletal muscle fiber hypereosinophilia (*), in horse skeletal muscle at 72 h. H&E. 200×.

Figure 5.

Myofiber loss of striation and sarcoplasmic fragmentation in semitendinosus skeletal muscle of 12 horses held at 22°C over 72 h postmortem. A. Each point is the average field involvement of the criterion in 5 fields at 200× magnification at each time (0 = absent, 1 = 1–20%, 2 = 21–40%, 3 = 41–60%), p = 0.003. B. Each point is the average field involvement of the criterion in 5 fields at 200× magnification at each time (0 = absent, 1 = 1–20%), p = 0.046. C. Loss of striation (*) and sarcoplasmic fragmentation (arrow) in horse skeletal muscle at 72 h. H&E. 200×.

For skeletal muscle maintained at 8°C, myofiber eosinophilia increased linearly (p ≤ 0.0001) and loss of striation (Suppl. Table 1) initially increased, peaked at T48h, and then decreased slowly, consistent with a significant parabolic correlation with time (p = 0.010). For all remaining criteria, there was no significant trend over time.

The criterion of segmental hypercontraction of skeletal muscle fibers or large-diameter dark fibers was absent in all examined sections of muscle from all horses at all times.

Brain

For brain maintained at 22°C, neuronal nuclear swelling decreased initially over time, with its lowest point occurring at T36h, before increasing (Suppl. Fig. 5A). This was statistically significant (p = 0.005), in a decreasing parabolic relationship with time. Vacuolation of the neuropil (Suppl. Fig. 5B) occurred in a significant (p = 0.004) nonlinear trend with time, increasing initially and peaking at T24h, and then decreasing, and was best fit as the square root over time. Hence, nuclei initially shrink then swell, while at the same time, vacuolation of the neuropil occurs and then regresses. Although these criteria were statistically significant, the percentage of field involved was 20–40%.

At 8°C, neuronal pyknosis (p = 0.039) as well as Nissl substance dissolution (p = 0.012) increased significantly as a linear trend over time (Suppl. Fig. 5C, 5D). Neuronal cytoplasmic swelling (Suppl. Fig. 5E) exhibited a significantly decreasing (p = 0.014) linear relationship with time, suggesting continuous shrinkage. Vacuolation of neuropil (Suppl. Fig. 5F) exhibited a nonlinear trend, with the best fit line corresponding to a square-root relationship with time (p = 0.012). Although these criteria were statistically significant, the percentage of field involved was 24–40%.

Neuronal cytoplasmic vacuolation was absent in all examined sections of brain from all horses at all times. We found that sex and age (young vs. adult) were not significant variables.

Discussion

We observed the most consistent and significant changes at 22°C in liver and muscle tissue, with criteria observed in >80% of the fields over time. Brain was the least predictable at either temperature. Most of the significant criteria showed linear correlations with time, whereas others demonstrated a significant parabolic, square-root, or exponential trend. For most of the nonlinear relationships, limited field involvement was observed, making these parameters unreliable. Occasionally, outliers were reported at single times, which were considered incidental and independent of autolysis. We observed that all histologic changes started early after death and evolved with time; no criteria appeared with delay at a later time.

In liver, individualization at 22°C changed the most extensively, becoming apparent in ≥80% of the fields examined at 72 h. This criterion was followed by bile duct epithelium separation from the basement membrane, and within bile duct epithelial pyknosis. In skeletal muscle, loss of striation at 22°C occurred in up to 60% of the fields examined by 72 h.

In a study¹⁶ of the sequence and rate of autolytic changes of hepatocytes and bile ducts over 48 h in guinea pig livers maintained at 22°C, hepatocellular cytoplasmic basophilia was lost completely by 3 h, with the cytoplasm staining more eosinophilic. No changes in hepatocyte basophilia were statistically significant at either temperature in our examined equine livers. Guinea pig hepatocyte cytoplasmic vacuolation and granularity increased progressively from 9 h through 36 h, when in horses, vacuolation was only significant at 8°C and observed as early as 1 h after death, and was evolving following a square-root relation with time over 72 h. Nuclear fading and chromatin margination was apparent at 9 h, and after 48 h, ~25% of the guinea pig hepatocyte nuclei had lysed. Hepatocyte chromatin margination in horses was not statistically significant at either temperature over 72 h.

Individualization of hepatocytes following dissociation of single or groups of hepatocytes was first noticed at 6 h and was most extensive at 48 h, when 25% of hepatocytes were affected in guinea pigs. Individualization of equine hepatic plates was observed as early as 1 h after death and was most extensive at 72 h, when up to 80% of hepatocytes were affected. In portal areas, the separation of bile duct epithelium from the basement membrane was first observed at 18 h in guinea pigs and was present in most bile ducts by 24 h. Bile duct epithelium separation from the basement membrane in horses was the only criterion observed with statistical significance at both temperatures, as early as 1 h postmortem and was most significant by 72 h at 22°C, affecting up to 30% of the bile ducts. Chromatin margination of guinea pig bile duct epithelial nuclei was first seen at 6 h, and by 48 h most bile duct epithelial cell nuclei were affected; chromatin margination in bile duct epithelial nuclei was not significant at either temperature in horses.

Another study¹⁸ described progressive hepatocellular pyknosis as well as hepatocyte individualization over 6 h of postmortem autolysis in the rat liver, which is similar to the progressive hepatocyte individualization seen in our study. Significantly contrasting results were seen in dog liver degradation.¹ Bile duct epithelium detachment from the basement membrane occurred at day 3 in dogs but was detected after 1 h in our horses. By day 7, most canine hepatocyte nuclei were autolyzed, and the most significant hepatocyte autolysis was observed at 3 wk. Although our study did not extend past 72 h, one could speculate that, by 3 wk, hepatocytes would also show the most significant autolysis. This illustrates that liver changes are clearly not comparable between species, and criteria should be established on a species-by-species basis. Those differences may be the result of differences in the size of animals, skin thickness, and hair coat, which affects the body cooling rate.

Autolytic changes occurring in skeletal muscle are less well studied. In one of the first studies, human skeletal muscle tissue was electrically stimulated after death⁷ to affect changes over time. In another study, changes were studied in canine skeletal muscle traumatized after death followed by exposure to sea water during the decomposition process.¹⁷ Changes consisted of interruption of individual skeletal muscle fiber continuity with loss of cross-striation (rupture of vital fibers), multiple interruptions of fiber continuity with formation of separate groups of sarcoplasm with loss of cross-striation (segmental disintegration of fibers), and disc-like separation of sarcoplasm with interruption of fiber continuity (discoid disintegration of fibers). In comparison, we observed the most significant changes in myofibers in horses at 22°C, and loss of striation was the only criterion observed significantly at both temperatures.

In general, most of the changes that we observed in the cerebrum were subtle and were not present in large percentages of the fields examined. Brain autolysis in humans showed neuronal Nissl substance dissolution, and karyorrhexis progressively increased from 5 to 22 h postmortem.^12,14,15 Nissl substance dissolution in horses was only significant at 8°C, was seen as early as 1 h, and increased linearly until 72 h. Neuronal cytoplasmic vacuolation, basophilic cytoplasmic staining, and nuclear and cytoplasmic swelling were also considered to be autolytic changes in humans. In horses, neuronal nuclear swelling was only significant at 22°C; cytoplasmic swelling was only significant at 8°C, and neuronal cytoplasmic vacuolation was not statistically significant at either temperature. Autolytic changes in the human brain were first discernable as swelling of the neuronal nucleus and cytoplasm with increasing chromatolysis and liquefaction of the cytoplasm that may or may not involve the nucleus.¹⁴ These initial changes appeared at different times after death,¹¹ with observations occurring as early as 30 min or as late at 3 h postmortem. Such chronology of events was not observed in equine brains; all autolytic changes in the brain were subtle, and no single most significant criterion stood out as a major marker of autolysis.

We observed a striking difference between the intensity and distribution of autolysis between the tissues kept at 22°C compared to the ones kept at 8°C (Suppl. Table 1). This was particularly evident in muscle tissue in which most criteria were significant at 22°C and not at 8°C, and in the case of loss of striation, its highest field representation was 60% at 22°C, and was only 10% at 8°C. Similar observations were made for the criteria in liver tissue, in which bile duct separation from the basement membrane was seen in 30% of the fields at 22°C, and in only 12% of the fields at 8°C. Hence, autolysis was both slowed with cooling and progressed as a different trend over time. In the liver, hepatocellular individualization, vacuolation, and bile duct epithelium pyknosis changed the most between temperatures and were only significant at 22°C. In brain and skeletal muscle, pyknosis, Nissl substance dissolution, neuronal nuclear and cytoplasmic swelling, interruption of myofiber continuity, and sarcoplasmic fragmentation of skeletal muscle cytoplasm were the most variable between temperatures. Some changes were not present at all or only present at 22°C by 72 h after death, and these included hepatocellular chromatin margination and basophilia, and margination of chromatin in bile duct epithelium nuclei.

Although the inter-horse variation could have been reduced by analyzing samples from the same horse for all times, a study involving separate animals was necessary to establish intra-horse variation based on age and sex that could be applied to future statistical models using the criteria found most reliable in our study.

Separation of tissues from each carcass was one limitation of our study; thus, our data do not reflect field conditions. Given the size of the horses, storage of 12 carcasses was not feasible. Most of the criteria that we studied could foreseeably be less reliable when analyzed in the context of a full carcass for several reasons. For instance, it would be expected that maintenance of vascular connections between organs would allow migration of gut microbiota to liver, which would accelerate putrefaction. Organs within the body would likely maintain higher temperature initially, allowing longer protein and enzyme activity of any process of degradation and therefore increase the rate of autolysis. The chosen temperatures in our study were 22°C and 8°C, which were the autopsy room and walk-in cooler temperatures.

Our equine candidates were selected based on their overall good health, adequacy of nutritional condition, and with a wide age range. With increasing age, there may be underlying conditions that our study did not account for. These may include geriatric (suboptimal vascular circulation) or neoplastic conditions that alter metabolic rates affecting lysosomal protein degradation during autolysis. Variation in body temperature could also play a major role in influencing protein or enzyme activity, whereby patients dying with fever would see an increased progression of autolysis versus patients dying in hypothermia. These factors may all affect decomposition rate and, if there is antemortem tissue necrosis, autolysis would be difficult to assess accurately. Gross postmortem analysis should allow one to discern necrosis from autolysis based on lesion location and distribution. Autolysis will affect the entire body, and an entire organ more uniformly, which microscopically will be seen in an entire section of tissue.

Given that our data were analyzed as non-repeated measurements, only correlation of trends with time could be established. This design and analysis lack confidence intervals (CIs) for the measures of central tendency and predictive value. Therefore, the results of our study are not ready to be used to estimate the postmortem interval in field cases. There may be a rough indication of PMI based on the extent of distribution in observed fields of each criterion. By increasing the number of subjects with repeated measurements of these significant criteria for autolysis, it could be possible to establish regression equations with CIs for the mean and possibly predictive values. If this is possible, then histologic criteria could be used to estimate the PMI as well as a 95% CI for the estimate of the time of death.

Taking the results from our study and what is known from the literature, it seems that criteria of autolysis should be established separately for each species. Multiple temperature levels with longer times could provide data that lay the groundwork for establishing PMI criteria that are more objective than current parameters such as body temperature.⁴ Investigation of additional tissues will provide an extensive and detailed picture of how autolysis affects tissues after death and could possibly be used to estimate the PMI.