The first successful liver transplant surgery was performed in 1963 by the pioneering surgeon Thomas Starzl. However it was not until the advent of cyclosporine and its introduction into clinical practice in the late 1970s and 1980s that adequate 1-year survival could be achieved and liver transplant became truly viable as a therapeutic intervention (Starzl et al. 1981). Initially, advanced HCC was thought to be a good indication for liver transplant. Although the volume of liver transplants performed in the 1980s for the indication of HCC varied by center, in at least one German transplant program, 37 % of patients underwent transplant for the primary indication of HCC (Scharschmidt 1984).

Unfortunately, early experiences with liver transplantation for HCC were notable for poor survival rates. For example, a 1984 multicenter analysis reported an overall 1-year posttransplant survival on the order of 40 % for HCC patients (see Fig. 3) (Scharschmidt 1984). Rates of recurrent HCC after transplant were high, ranging from 53 % to as high as 82 % (Koneru et al. 1988; Iwatsuki et al. 1985).

Based on these early experiences, in 1984, a National Institutes of Health (NIH) conference on liver transplant was convened. The consensus statement from this conference concluded that although survival data were limited at the time, “results to date indicate a strong likelihood of recurrence of the malignancy” but that “the procedure may achieve significant palliation” (Anonymous 1984). Indeed, in a 1985 paper, Starzl’s group wrote that “it has been tempting during the acquisition of this experience to conclude that liver replacement for malignant hepatic neoplasms is conceptually unsound, except for fibrolamellar hepatomas, and to abandon such efforts” (Iwatsuki et al. 1985). The authors even went so far as to speculate that total body irradiation and chemotherapy might be necessary adjuvant therapies to make liver transplant a curative intervention for HCC (Iwatsuki et al. 1985). Likewise, in a 1987 commentary, Bismuth noted “it is in this group [primary liver cancer] that the long-term results for transplantation are poorest due to the appearance of recurrent disease” (Bismuth 1987). The increasing demand for donor organs for use in patients with superior anticipated outcomes soon precluded the use of liver transplantation with palliative intent.

Throughout the 1980s and early 1990s, additional reports confirmed the observation that patients transplanted for HCC did poorly, with high rates of recurrent HCC and limited overall survival (O’Grady et al. 1988; Jenkins et al. 1989; Ismail et al. 1990). In the most heartening of these early reports, Haug et al. reported in a 1992 paper a 25 % rate of recurrent HCC in patients surviving at least 3 months after transplant, which was good compared to other reports of the time, but which is still high by current standards (Haug et al. 1992). Although overall outcomes for liver transplant patients in general were improving during this time period, due to advances in surgical technique and immunosuppression, outcomes for HCC patients remained poor in comparison to those without HCC (see Fig. 4).

In this era, attempts to stratify patients into groups at acceptably low risk for recurrent HCC versus unacceptably high risk met with limited success. For example, some studies examined using size and number of HCC masses as a predictor of outcome, but did not find this to be helpful (O’Grady et al. 1988; Olthoff et al. 1990). These early findings, which are contrary to our current understanding of the relationship between tumor burden and risk for recurrent HCC, were likely in large part due to the fact that so many of the patients included in these analyses had tumor that was very advanced by current standards. Because so many patients in these early studies already had quite significant tumor burden, well beyond the Milan Criteria, and were already at quite high risk for recurrent HCC posttransplant, quantification of tumor within this group was of limited utility. The one exception was that patients found to have HCC as an incidental finding on explant, as compared to having a known HCC diagnosis prior to transplant, did substantially better in terms of HCC recurrence and overall survival (Iwatsuki et al. 1985; Jenkins et al. 1989) (see Fig. 5). Indeed, in a 1989 paper, Ringe et al. noted that “at present, the factors playing a major prognostic role – being either especially advantageous or disadvantageous – are virtually unknown, and it is impossible to predict those patients who are most likely to have prolonged survival without tumor recurrence. Obviously there are only two known exceptions to the generally poor prognosis in cancer patients: incidental hepatomas arising in livers with other diseases, and the fibrolamellar variant of hepatocellular carcinoma” (Ringe et al. 1989). Today this observation can be understood in terms of “incidental hepatoma” serving as a surrogate for HCC staged comfortably within the Milan Criteria.

Interpretation of these early data is confounded by several factors. First, many of these early papers combined HCC patients together with cholangiocarcinoma patients and even with cases of metastatic cancer of non-hepatic primary. Second, quantification of tumor burden, if reported at all, was done in a rudimentary fashion. Third, many of the patients undergoing transplant with HCC were non-cirrhotic, in contrast to the situation at present in the United States. Additionally, some patients in these early series underwent embolization therapy, further complicating understanding of tumor burden (Moreno Gonzalez et al. 1992; Pichlmayr 1988; Olthoff et al. 1990). However, the clear historical lesson from these early experiences with liver transplantation for HCC is that the availability of liver transplantation as a therapeutic option for HCC patients was in jeopardy. The option of liver transplantation was in danger of being closed to HCC patients unless some method could be developed to identify at least a subset of HCC patients with a reasonable expectation of adequate long-term survival to justify the use of scarce donor organs.

Initial observations that small incidental HCCs identified on explant carried less ominous prognosis in comparison to known large HCC led to a more refined appreciation of the relationship between tumor burden and risk of posttransplant recurrence in the ensuing period. However, before reviewing these data in detail in the following section, it is important to note the absence of any one uniform staging system used to quantify the degree of HCC tumor burden.

In fact, a multitude of HCC staging systems is available. Although pathologists do routinely use the standardized “TNM” (tumor-node-metastasis) system for describing findings in surgical specimens, there is no such uniform standard used by hepatologists, surgeons, and clinical researchers. Therefore, studies assessing the relationship between tumor burden and clinical outcomes use the language of the TNM system, the BCLC, Cancer of the Liver Italian Program (CLIP), Okuda, and others. These staging systems generally incorporate measures of the degree of underlying liver disease in addition to quantification of tumor burden. It is beyond the scope of this chapter to detail these various alternative staging systems, but suffices to state that analysis of the relationship between tumor stage and clinical outcome has been made more complex by this proliferation of competing clinical staging systems. Additionally, the TNM system itself has undergone a series of significant changes over time, notably including the introduction of TNM staging for primary hepatobiliary cancer in the 2nd edition of the American Joint Committee on Cancer (AJCC) staging manual, and a major change to the staging system in the 6th edition in 2002. The evolution of the TNM staging system as defined by the AJCC from its origins in 1977 to the current 7th edition released in 2010 is detailed below (Table 1).

In the early editions of the AJCC TNM staging system, T1 stage indicated a small solitary tumor without vascular invasion, whereas after 2002 in the 6th and 7th editions, T1 took on an entirely different meaning, encompassing solitary tumors of any size as long as vascular invasion was not present. It is important for the present-day reader to keep in mind the changing meaning of TNM staging over time when considering publications assessing the relationship between tumor burden and clinical outcomes. It is of particular note that in none of the various versions of the AJCC TNM system did a particular TNM stage correlate with the Milan Criteria. For example, at the time of publication of Mazzaferro’s 1996 paper, patients who fell within the Milan Criteria could have tumor staged by then-current 4th edition of TNM anywhere from T1 to T4, and patients with HCC beyond Milan could be staged by TNM anywhere from T2 to T4. Similarly using the now-current 7th edition of TNM, a patient falling within the Milan Criteria could have HCC staged by TNM either T1 or T2, and patients beyond the Milan Criteria could be staged by TNM anywhere from T1 to T4.

In light of the adoption of the Milan Criteria, a new staging system initially entitled the “American Liver Tumor Study Group-Modified TNM Staging System” was defined (see Table 2). Critically, this new staging system was developed for clinicians and was discordant with the AJCC TNM system used by pathologists. The “American Liver Tumor Study Group-Modified TNM Staging System” has since been variously referred to as “UNOS stage,” “Milan stage,” and/or “clinical stage” and is the basis of pretransplant radiographic staging of HCC under current UNOS/OPTN policy (Befeler et al. 2005; Schlansky et al. 2015).

In summary, the reader is cautioned when encountering “T1,” “T2,” etc., phraseology in the literature cited in this chapter. The meaning of such classification is dependent not only on which staging system is employed but, in the case of the AJCC TNM system, also on the edition of the staging manual.

Although some earlier studies, as noted above, failed to detect an association between quantification of tumor burden and clinical outcomes, in the late 1980s and through the 1990s, a number of studies did begin to detect a correlation between clinical outcome and TNM stage of HCC assessed on the explanted liver (Pichlmayr et al. 1995; Iwatsuki et al. 1991; Ringe et al. 1991; Selby et al. 1995). Higher degrees of tumor burden were associated with worse clinical outcomes.

Independent of studies using the established TNM system, other groups assessed novel cutoffs in terms of number and size of HCC masses and similarly reported better clinical outcomes for patients with more limited tumor burden. For example, Yokoyama et al. reported that the rate of posttransplant recurrence of HCC was 64 % in all HCC patients, but was reduced to 37 % in those patients with maximal tumor diameter of 5 cm or less (Yokoyama et al. 1990). McPeake et al. observed significantly better overall survival for patients classified into a low-risk group characterized by having solitary tumor no more than 4 cm, as compared to a high-risk group with multifocal disease or solitary tumor in excess of 8 cm and also as compared to the intermediate risk group comprised of the remaining patients (McPeake et al. 1993) (see Fig. 6). Similarly, Bismuth et al. observed a trend toward better outcomes for a low-risk group of patients with tumors less than 3 cm, as compared to a moderate risk with tumor between 3 and 5 cm, and a high-risk group with tumor in excess of 5 cm (Bismuth et al. 1993). Although this relatively small study involving 60 HCC patients undergoing transplant was not powered to demonstrate statistical significance of this three-tiered stratification of risk, they were able to demonstrate with statistical significance that patients with either one or two tumors no more than 3 cm in diameter had better outcomes in comparison to those with higher degrees of tumor burden (Bismuth et al. 1993). Tan et al., in a 1995 paper, reported that when transplant was restricted to patients with tumors less than 8 cm in diameter, recurrence rates were held to 15 %, in a small patient cohort with a minimum of 18 months of clinical follow-up (Tan et al. 1995).

In the context of growing evidence of a relationship between tumor burdens measured in terms of number and size of HCC masses, Figueras et al. reported in a 1997 paper that outcomes of patients with HCC can equal those transplanted without HCC (Figueras et al. 1997). In this study, 89 % of patients with HCC had maximal tumor diameter less than 5 cm and 66 % had either one or two masses. Actuarial survival was equivalent between this group of patients and a group undergoing transplant without HCC (Fig. 7). Of note, the great majority of HCC patients in this study did undergo trans-arterial chemoembolization prior to transplant.