Predictive value of frozen-section analysis in the histological assessment of steatosis before liver transplantation



Histological quality assessment of donated livers is a key factor for extending the cadaveric donor pool for liver transplantation. We retrospectively compared frozen-section analysis with routine histological permanent slides and the outcomes of grafts in liver biopsies from 294 candidate donors. The κ concordance coefficient of agreement between frozen-section analysis and routine histological analysis was very good for macrosteatosis (κ = 0.934), microsteatosis (κ = 0.828), and total steatosis (κ = 0.814). The correlation between the mean amounts of macrosteatosis, microsteatosis, and total steatosis in frozen and permanent sections was also significant (P < 0.001, Spearman's test). Macrosteatosis and microsteatosis were overestimated to >30% in 4 of 32 cases (12.5%) and in 23 of 62 cases (37.1%), respectively. The only 2 histological parameters of frozen sections able to predict graft dysfunction within 7 days of transplantation were macrosteatosis and total steatosis (P = 0.018 and P = 0.015, respectively, Mann-Whitney test). None of the other histopathological features evaluated in frozen sections, including portal inflammation, lobular necrosis, myointimal thickening, biliocyte regression, cholestasis, hepatocellular polymorphism, lipofuscin storage, and fibrous septa, were significantly correlated with the graft outcome. The frozen-section histological evaluation of biopsies from cadaveric liver donors is an accurate, time-effective, and predictive method for the assessment of graft suitability. Liver Transpl 15:1821–1825, 2009. © 2010 AASLD.

Quality assessment of donated livers is a key factor for extending the donor pool for liver transplantation. Severe liver steatosis, with particular reference to macrovesicular steatosis, represents the main contraindication to the use of apparently healthy livers because of the mechanical obstacle of fat droplets to the blood flow at the time of graft liver reperfusion.1 The eligibility of donated livers for transplantation is generally assessed by a combined clinical ultrasound, surgical macroscopic, and pathological microscopic evaluation. There is no current agreement among pathologists on the histological method of choice for pretransplant liver donor evaluation. Most of the previous studies on the quality assessment of donated liver tissue were performed with a routine histopathological procedure on organs from both living and cadaveric donors.2–4 In other reports on large series of pretransplant liver biopsies, the method used for the histological analysis was not specified.5, 6 The frozen-section approach implies lower histopathological definition but allows a rapid histological examination. It also appears to be preferable for cadaveric donors because it implies a substantial reduction of the cold ischemia time in comparison with routine histopathological processing.7 The main limitation ascribed to the frozen-section procedure is represented by the risk of underestimating macrosteatosis while overestimating microsteatosis due to the presence of water droplets entrapped in hepatocytes at the time of tissue freezing. Nonetheless, we recently reported that the frozen-section evaluation of pretransplant liver biopsies led to the successful utilization of livers from donors > 60 years old.8 Few reports thus far have compared the influence of using these 2 histological techniques on the final outcomes of liver grafts from cadaveric donors.9

Here we retrospectively reviewed 294 liver biopsies from cadaveric donors and compared the histopathological evaluations of frozen sections and corresponding permanent sections. The histology of frozen sections was correlated with the histology of corresponding permanent sections and the outcomes of grafts at different time points.


EPGF, early poor graft function; OLT, orthotopic liver transplantation; PGD, primary graft dysfunction.


Patient Population

We retrospectively collected 294 liver biopsy specimens taken from 141 female cadaveric liver donors and 153 male cadaveric liver donors (mean age = 67.16 ± 11.5; range = 18-86) between January 2001 and December 2005 in the 15 intensive care units of the Italian Emilia-Romagna Region. Informed consent for the donation of organs and tissues was obtained from the closest donor relative or legal representative according to Italian regulations.

All candidate donors underwent the screening protocol for infectious and neoplastic diseases currently applied in our institution as previously described.10 The protocol included an ultrasound liver examination and an evaluation of serum bilirubin and aminotransferases. Wedge and fine-needle liver biopsies were taken from each donor before transplantation at the time of vascular clamping.8 Biopsies were sent in gauze soaked with a saline solution to the regional, centralized 24-hour pathology service in Bologna for frozen-section evaluation. At least two 4-μm sections were cut from each sample. After routine hematoxylin and eosin staining, an extemporary pathology report was produced, including the evaluation of the following histopathological features: the extent of macrosteatosis, microsteatosis, and total steatosis (semiquantitatively scored as the percentage of hepatocytes with microdroplets and/or macrodroplets of fat over the total number of hepatocytes); the intensity of portal inflammation (mild, moderate, or severe); the extent of myointimal thickening (mild or severe); the presence of biliocyte regression; the presence of lobular necrosis; the intensity of cholestasis (mild, moderate, or severe); the presence of hepatocellular polymorphism; the presence of lipofuscin storage; and the presence of fibrous septa. On the basis of the combined evaluation of the histopathological features of frozen sections and the gross surgical inspection, the livers were accepted or denied for transplantation. Macrovesicular steatosis was defined as the accumulation of lipid droplets of various sizes within hepatocytes in which the cell nuclei were displaced peripherally. Microvesicular steatosis was defined as the presence of foamy lipid droplets in hepatocytes with centrally located nuclei.11 A general threshold of 30% macrosteatosis was used as the histological criterion for selection. Permanent histological sections were routinely prepared for each biopsy after fixation and were not used for any decision during the transplantation process. Permanent slides were reviewed for the specific scope of this study by 2 expert liver pathologists (M.F. and F.V.) blinded to the original frozen-section evaluation.

Organs were allocated in the Emilia-Romagna Region according to a modified Model for End-Stage Liver Disease system as previously described.12 Orthotopic liver transplantation (OLT) was performed in the 2 transplant centers of the Emilia-Romagna Region (University of Bologna and University of Modena). The outcome after OLT was available for 122 patients who underwent transplantation at the University of Bologna. Histopathological features were correlated with the function of the graft at 7 days, 3 months, 6 months, and 1 year after the first OLT procedure. According to the criteria suggested by Tekin et al.,13 primary graft dysfunction (PGD) was classified as poor function of the graft leading either to the death of the recipient or to a second transplant or as levels of aspartate aminotransferases > 1.500 U/L and a prothrombin time > 20 seconds within 7 days after OLT.

Statistical Analysis

Data are reported as means and standard deviations, ranges, and frequencies. Groups of variables were compared with Mann-Whitney, Spearman, and Pearson's chi-square tests. The interobserver agreement between frozen and permanent sections of the same biopsies was assessed with the κ statistical method. A 2-tailed P value less than 0.05 was considered statistically significant. Data analysis was performed with SPSS for Windows (version 13).


The Assessments of Liver Steatosis in Frozen and Permanent Sections Are Nearly Equivalent

Histopathological data on liver steatosis in frozen and permanent sections were available for 279 of 294 biopsies (95%). In frozen-section analysis, the mean degree of macrosteatosis was 14.07% ± 17.31% (range = 0-90), the mean degree of microsteatosis was 19.76% ± 22.36% (range = 0-100), and the mean degree of total (combined) macrosteatosis and microsteatosis was 33.73% ± 31.65% (range = 0-170). In the permanent sections of the same biopsies, the mean degree of macrosteatosis was 13.69% ± 17.46% (range = 0-100), the mean degree of microsteatosis was 15.75% ± 17.94% (range = 0-80), and the mean degree of total (combined) macrosteatosis and microsteatosis was 29.44% ± 27.84% (range = 0-110; Fig. 1). Spearman's correlation coefficient for the histological assessment of steatosis in frozen and permanent sections was highly significant (P < 0.001) for macrosteatosis, microsteatosis, and total steatosis.

Figure 1.

Comparison of (A,B) microsteatosis and (C,D) macrosteatosis in (A,C) frozen and (B,D) permanent sections. The extent of microsteatosis was <30%, whereas the extent of macrosteatosis was >30%. The correlation between the mean amounts of all types of steatosis in the frozen and permanent sections was highly significant (P < 0.001, Spearman's test).

The coefficient of agreement (κ concordance coefficient) between the group of pathologists that evaluated the frozen sections at the time of OLT and the 2 pathologist that blindly reviewed the permanent slides was always very good: κ = 0.934 for macrosteatosis, κ = 0.828 for microsteatosis, and κ = 0.814 for total microsteatosis and macrosteatosis.

To analyze the macrosteatosis data according to the 30% cutoff used for the acceptance/exclusion of candidate livers, we stratified the percentage of steatosis into the 3 following groups: <10% macrosteatosis, between 10% and 30% macrosteatosis, and >30% macrosteatosis. In the frozen-section analysis, the stratification of macrosteatosis was as follows: 181 (64.9%) with <10% macrosteatosis, 66 (23.6%) with 10% to 30% macrosteatosis, and 32 (11.5%) with >30% macrosteatosis. In the corresponding permanent sections, 180 cases (64.5%) had <10% macrosteatosis, 70 (25.1%) had 10% to 30% macrosteatosis, and 29 (10.4%) had >30% macrosteatosis. In particular, 4 of 32 cases (12.5%) with >30% macrosteatosis in the frozen sections were underscored to 10% to 30% macrosteatosis in the permanent-section evaluation. Conversely, just 1 of the 66 cases (1.5%) graded with 10% to 30% macrosteatosis in the frozen sections was overscored to >30% macrosteatosis in the permanent sections.

When we looked at the microsteatosis, the stratification was as follows: 149 (53.4%) with <10% microsteatosis, 68 (24.4%) with 10% to 30% microsteatosis, and 62 (22.2%) with >30% microsteatosis in the frozen sections versus 163 (58.4%) with <10% microsteatosis, 76 (27.3%) with 10% to 30% microsteatosis, and 40 (14.3%) with >30% microsteatosis in the permanent sections. In particular, 23 of 62 cases (37.1%) with >30% microsteatosis in the frozen sections were underscored to <30% microsteatosis in the permanent-section evaluation, whereas just 1 (0.7%) of the 149 cases deemed to have <30% microsteatosis in the frozen-section evaluation was overscored to >30% microsteatosis in the permanent-section evaluation. The κ concordance coefficients for the assessment of macrosteatosis and microsteatosis in the frozen and permanent sections for these 3 groups were very good (κ = 0.916) and good (κ = 0.799), respectively.

Thus, the histological evaluation of all types of liver steatosis seems to be equivalently reliable in frozen sections and corresponding permanent slides.

The Amount of Macrosteatosis and Total Steatosis in Pretransplant Liver Biopsy Specimens Predicts PGD but Not Long-Term Loss of the Graft

Data regarding graft outcomes at 7 days, 3 months, 6 months, and 1 year after OLT were available for 122 of the 176 patients (69%) transplanted at the Bologna center. The mean follow-up after OLT was 1022 ± 620.5 days. The mean survival of the graft was 947.68 ± 631.63 days (range = 0-2276), and 39 of 122 grafts (31%) were lost at the time of the last follow-up. In particular, 9 patients (7%) experienced PGD within 7 days. Other causes of graft loss included hepatitis C virus–related hepatitis recurrence (12 cases, 30.8%), sepsis (9 cases, 23%), neoplasia (3 cases, 7.7%), surgical complications (2 cases, 5.1%), chronic rejection (1 case, 2.7%), and nonassessed (3 cases, 7.7%).

In the 9 patients who experienced PGD, the mean macrosteatosis, microsteatosis, and total steatosis at pre-OLT biopsy were 17.78 ± 11.49, 28.33 ± 23.18, and 46.11 ± 26.07, respectively, whereas among the 113 patients with a functional graft at 7 days, the corresponding mean values were 9.85 ± 10.94, 15.3 ± 19.98, and 25.15 ± 26.64 (Table 1). The only 2 histological parameters predictive of PGD after OLT were macrosteatosis and total steatosis (P = 0.018 and P = 0.015, respectively, Mann-Whitney test; Table 1). Conversely, no statistical correlation was found between the amounts of microsteatosis, macrosteatosis, and total steatosis at the time of pretransplant biopsy and the function of the graft at 3 months, 6 months, and 1 year after OLT (data not shown).

Table 1. Differences Between the Mean Macrosteatosis, Microsteatosis, and Total Steatosis in the Pretransplant Biopsy Specimens of Patients With or Without Experience of PGD
 No EPGFEPGFMann-Whitney P Value
  1. Abbreviation: EPGF, early poor graft function.

Mean macrosteatosis9.8517.780.018
Mean microsteatosis15.328.330.074
Mean total steatosis25.1546.110.015

None of the other histopathological features evaluated in the same frozen-section biopsies and outlined in Table 2 were significantly correlated with either short- or long-term graft outcomes.

Table 2. Correlations Between Histopathological Features and Their Frequency and the Occurrence of PGD After Liver Transplantation
  Frequency%EPGF P Value
  • Abbreviation: EPGF, early poor graft function.

  • *

    Spearman's test.

  • Pearson's chi-square test.

Total steatosis<10%6351.6 
Portal inflammationAbsent6235.4 
Myointimal thickeningAbsent11364.6 
Biliocyte regressionAbsent13979.4 
Lobular necrosisAbsent14582.9 
Hepatocellular polymorphismAbsent15588.6 
Lipofuscin storageAbsent16493.7 
Fibrous septaAbsent16494.3 


The histological evaluation of donor liver quality is essential for the success of OLT. Among the pathological features, liver steatosis (particularly macrovesicular steatosis) represents the major cause of exclusion from donation or early graft failure.1, 14 When we are dealing with cadaveric donors, the key point for the preservation of liver grafts is the minimization of the cold ischemia times. The optimal interval between donor liver clamping and reperfusion is approximately 4 hours and frequently includes transport from a peripheral location to a centralized transplant center. Frozen-section evaluation of the donor liver biopsy represents the best time-saving methodological approach to complying with these short terms because routine histological processing itself generally requires no less than 3 hours.

Our data suggest that, at least in our setting, the histological evaluations of mean liver macrosteatosis, microsteatosis, and total steatosis in frozen sections and corresponding permanent sections virtually overlap. A limitation generally ascribed to the use of frozen sections is the overestimation of steatosis (particularly microsteatosis) due to the technical artifact of water droplets trapped during the freezing procedure. According to some reports, this might lead to the wrong estimate of steatosis over 30% and, therefore, to the exclusion of suitable organs from donation. In our series, we stratified the cases according to the 30% cutoff, and we recorded an overestimation of macrosteatosis in just 4 of 279 biopsies (1.4%). We realize that the rate of overestimation for microsteatosis over 30% in our frozen-section series was higher (8.2%). However, the score of microsteatosis is generally not used as a criterion for donor exclusion. Furthermore, unlike macrosteatosis, the extent of microsteatosis did not correlate with the outcome of the graft (as discussed later). We can therefore conclude that frozen-section analysis might be considered the approach of choice for the evaluation of liver biopsies from cadaveric donors. In this view, particular attention should be paid to the shipping of the biopsy specimens to the pathology service, the avoidance of excessive soaking of the samples in a saline solution or other media, and the transportation of the biopsy specimens in a dry environment such as dry gauze or empty vials. Another technically useful expedient for excluding artifacts might be a relatively slow freezing process at −20° without the use of liquid nitrogen for snap freezing.

We also found that the rates of macrosteatosis and total steatosis at the time of transplantation were statistically able to predict the short-term graft outcome but not the long-term outcome. This is not unexpected in liver transplantation because of the high rate of graft loss in long-term follow-up due to different clinical conditions such as hepatitis recurrence, sepsis, and de novo tumor occurrence. It is also not surprising that the best histological predictors of primary nonfunction were the amounts of macrosteatosis and total steatosis but not the amount of microsteatosis. In fact, the big size of the fat droplets represents a mechanic obstacle that cannot be easily bypassed at the time of blood reperfusion. The exact size and average amount of macrosteatosis responsible for the development of poor graft function are not measurable and may be variable according to the specific characteristics of the donors and recipients. The general 30% cutoff used also in our institution represents the best compromise for this issue. Nonetheless, in our data, the mean degree of macrosteatosis in the patients who experienced PGD reached 17.8% and was approximately double that (9.8%) of the grafts that were functioning at 7 days. We can therefore conclude that the 30% cutoff for macrosteatosis should always be used for exclusion from donation. However, even with rates of macrosteatosis <30%, a thorough evaluation of the histological and macroscopic features of the liver during the surgical inspection should be made. In contrast, our results suggest that an extent of microsteatosis even >30% should not be considered a necessary criterion of exclusion from donation.

We also included in the present study an assessment of several other histological features in the pretransplant biopsy. These parameters belong to a histological checklist that is usually completed by our pathologists during the evaluation of graft eligibility. None of these features evaluated in frozen or permanent sections turned out to be correlated in our series with graft outcome at any time point. This finding might be explained by the preselection of all the candidate liver donors presenting in our region by means of ultrasound scanning and assessment of liver function blood tests. With this screening procedure, all grafts with major liver pathological conditions such as cirrhosis, diffuse necrosis, or severe cholestasis are discarded a priori. Therefore, we cannot exclude that these histological features could be more relevant in other settings, and thus this kind of checklist should be applied in any liver transplant–dedicated pathology service.

In conclusion, the frozen-section histological evaluation of biopsies from candidate cadaveric liver donors can provide transplantation teams with time-effective and predictive information about the eligibility of grafts.