Ethical approval
The experimental protocol was approved by the Danish Animal Research Committee, Copenhagen, Denmark (licence number: 2021-15-0201-00978). Animals received care in accordance with the US National Institutes of Health’s Guide for the Care and Use of Laboratory Animals15. The study is reported in accordance with ARRIVE guidelines16.
Healthy, 7–8-week-old male Wistar rats (Janvier Labs, Le Genest-Saint-Isle, France) with a mean preoperative weight of 270 g (range: 240–300 g) were housed in standard animal laboratories with a temperature maintained at 23 °C, an artificial 12-h light–dark cycle, and free access to food and water. The animals were kept in these facilities until the end of the experiment.
Experimental design
Sixty-eight male Wistar rats were randomized in blocks of two to either 90% PH (n = 44) or sham operation (n = 18) (Fig. 1). Each group was block randomized into three subgroups for euthanasia at 12 h (n = 10), 24 h (n = 14), or 48 h (n = 20) after surgery.
Animals presenting with a general distress score (GDS)17 ≥ 10 during day or GDS ≥ 6 at midnight were euthanised. Our previous publication provides a comprehensive description of the GDS evaluation18. Animals euthanised before planned evaluation were allocated to the “non-survivors” group and the subgroup closest to the actual time of euthanasia. Animals with a GDS < 10 at the planned time of euthanasia were allocated to the “survivors” group. A baseline group (n = 6) not undergoing laparotomy was also included.
Surgical procedure and tissue sampling
We previously described the surgical and anaestetic procedures in detail18. In brief, the surgical procedure was performed with inhalation anesthesia, using a mixture of oxygen (0.3 L/min), nitrous oxide (0.15 L/min), and sevoflurane (5%). 90% PH was performed by resecting the left lateral lobe, mediane lobe, right superior lobe, and right inferior lobe, leaving only the anterior caudate lobe and posterior caudate lobe intact. Sham-operated animals were subjected to laparotomy without PH.
The animals were observed continuously for the first three hours after surgery and then at least four times per day using a GDS. Two independent observers allocated a GDS value to each animal. Early euthanasia was performed if the GDS was ≥ 10 at any time during the day or ≥ 6 at midnight. Animals that died unexpectedly were autopsied to establish the cause of death.
At 12, 24, or 48 h, the animals were anesthetised again. Blood samples were collected from the heart by cannulation, and the liver remnant, consisting of the anterior and posterior caudate lobes, was harvested. Euthanasia was performed by cervical dislocation under anaesthesia.
Biochemical analyses
Blood samples were processed, snap-frozen in liquid nitrogen and stored at -80 °C until analysis. Alanine aminotransferase (ALT), alkaline phosphatase, haptoglobin, bilirubin, albumin, ammonia, creatinine, sodium, potassium and phosphate were measured using an immunoassay and clinical chemistry analyser (Siemens Healthcare Diagnostics Inc., Erlangen, Germany). The prothrombin-proconvertin ratio (PP ratio) was measured using Sysmex CS-2100i (Sysmex Corporation, Kobe, Japan), an automated blood coagulation analyser.
Liver weight and regeneration ratio
The pre-operatively estimated liver weight was calculated from the resected liver weight:
$$Estimated\, liver\, weight(preoperatively)= \frac{Resected\, liver\, weight}{Size\, of\, PH}\times 100$$
The change in liver weight was considered the regeneration ratio:
$$Regeneration\, ratio=\frac{\frac{Liver\, weight\left(euthanization\right)}{Body\, weight(euthanization)}}{\frac{Estimated\, liver\, weight\left(preoperatively\right)}{Body\, weight(preoperatively)}} \times 100$$
Stereology
Tissue preparation
The posterior caudate lobe was fixed in phosphate-buffered formalin for 24–48 h and cut into 2-mm thick parallel slices using a tissue slicer. Tissue slices were placed with the same side up and embedded in paraffin. From each paraffin-embedded block, three 3-µm thick sections were cut, providing systematic, uniformly random sampling sections for immunostaining and further analysis19.
Immunohistochemistry
Immunohistochemical staining of the serial sections was performed using an automated slide stainer (Benchmark Ultra; Ventana Medical Systems, Roche, Basel, Switzerland). All sections were deparaffinised and boiled in Ventana CC1 buffer (Ventana Medical Systems, Roche, Basel, Switzerland), pH 8, for heat-induced epitope retrieval. The sections were incubated for 32 min with two different antibodies. Ready-to-use monoclonal mouse anti-rat Ki-67 specific antibody (clone 30–9; Ventana Medical Systems, Roche, Basel, Switzerland) diluted at 1:20 was used as a proliferation marker, and anti-beta-catenin antibody (clone β-Catenin-1; Agilent Technologies, Santa Clara, CA, U.S.) diluted at 1:100 was used as a hepatocyte cell surface marker. The sections were counterstained with haematoxylin. An OptiView DAB Detection Kit (Roche, Basel, Switzerland) and UltraView Universal Alkaline Phosphatase Red Detection Kit (Roche, Basel, Switzerland) were then used for visualization of the bound antibodies.
Quantification
An investigator blinded to the section treatments analysed the sections using an Olympus BX50 microscope, modified for stereology using a motorized stage (Märzhäuser Wetzlar MFD, Wetzlar, Germany) and a digital camera (Olympus DP73, Olympus, Tokyo, Japan) connected to a computer running newCAST version 2020.08.4.9377 software (Visiopharm, Hørsholm, Denmark). The same investigator analysed all the sections.
Three thin serial sections were used to estimate the total number of Ki-67-positive-stained hepatocyte nuclei profiles per area of liver tissue. Microscopy was performed using a 60 × oil objective lens (NA 1.30). A 2-dimensional (2D) unbiased counting frame (65 × 45 μm2) was used to sample hepatocyte nuclei profiles, in, on average, 100 systematically uniformly and randomly selected fields of view per animal. Positive staining was defined as a counterstained red oval hepatocyte nucleus. In the 48-h survivor group, approximately 150 positive stained hepatocyte nuclei were counted per animal. Corners of the counting frame formed 4 test points. For every counting frame, the number of test points overlapping the liver tissue was counted.
Total count of Ki-67-positive hepatocyte nuclei per area (NA) was calculated using the following formula:
$${\mathrm{N}}_{\mathrm{A}}=\frac{\sum \mathrm{Q}}{\frac{\mathrm{a}}{\mathrm{p}} x \sum \mathrm{P}}$$
where ΣQ is the total count of Ki-67-positive hepatocyte nuclei profiles using the 2D unbiased counting frame in three sections, a/p is the area of the unbiased counting frame divided by four, and ΣP is the sum of the test points overlapping the liver tissue in the sections. The counting rules are illustrated in Fig. 2.
Ki-67 and beta-catenin-stained section from a 48-h survivor rat. Counting frames are displayed; green lines are inclusion lines and red lines are exclusion lines. The universal counting rule states that a positive hepatocyte profile is counted if the nucleus is significantly stained and if it is entirely within the counting frame, or if it touches an inclusion line and does not touch an exclusion line (green arrows). Positive nuclei touching an exclusion line are not counted (red arrow)19.
Hepatocyte volumes were estimated on hepatocytes with a beta-catenin-stained brown cell membrane using a 3D isotropic nucleator (Fig. 3). The nucleator was applied to each of the sampled hepatocyte profiles. The software calculated the volume of each sampled hepatocyte. For each section, mean hepatocyte volume was calculated before ratio-of-sums for the three serial sections was measured.
Ki-67 and beta-catenin-stained section from a 48-h survivor rat. Hepatocyte volumes are estimated using the principles of the 3D isotropic nucleator on a 3 µm-thick section. Counting frames are displayed as red exclusion lines and green inclusion lines. The universal sampling rule is described in Fig. 2. In this study, the objects of interest were hepatocytes, both Ki-67-positive and -negative, with a clear brown beta-catenin-stained cell membrane. The sampling unit was the hepatocyte nuclear profile. The software then generated two systematically random sampled test lines running through the nucleus of the hepatocyte. Intercept lengths of the cell membrane were marked. By using the distance along the two isotropic lines radiating from the nucleus to the boundary of the cell membrane, the software estimated the volume of the hepatocytes.
Statistical analyses
All variables were tested for normality using histograms and QQ-plots. When normality was violated, data were log-transformed to ensure normality, analyzed, and then back-transformed. The groups were compared using an analysis of variance, whereas an unpaired t-test was used for pairwise comparisons. Mean values are shown with their 95% confidence intervals (CIs). P-values of < 0.05 were considered significant. To evaluate intra-observer variance, the number of positive hepatocyte nuclei was first counted for all animals. Hereafter, the investigator randomly selected 8 sections from the 90% PH groups and counted the number of positive hepatocyte nuclei once again. The agreement between the first and second counts was calculated using intra-class correlation coefficients (two-way mixed model). For inter-observer variation in GDS measurements, Cohen’s weighted kappa was calculated for two raters. All statistical analyses were performed using Stata version 17.0 (StataCorp LLC, College Station, TX, U.S.).