East–West pathology agreement on precancerous liver lesions and early hepatocellular carcinoma #

The notion of cancer development as a multistep process, recognized in experimental liver carcinogenesis in rats,1 also applies to human hepatocellular carcinoma (HCC). This implies the occurrence of recognizable precancerous lesions before the development of overt malignancy. Years ago, studies of small nodular liver lesions resulted in a confusing plethora of terms, such that at the World Congresses of Gastroenterology in Los Angeles in 1994, an ad hoc International Working Party (IWP) agreed upon a standardized classification of nodular liver lesions with rationalized nomenclature.2 DN, dysplastic nodule; eWDHCC, early well-differentiated hepatocellular carcinoma; HCC, hepatocellular carcinoma; HGDN, high-grade dysplastic nodule; ICGHN, International Consensus Group for Hepatocellular Neoplasia; IWP, International Working Party; LGDN, low-grade dysplastic nodule. The precancerous lesions proposed by the IWP comprised microscopic and macroscopic alterations. Microscopic changes of less than 1 mm in diameter are referred to as dysplastic foci with either large or small cell change. Large cell change (previously large cell dysplasia) is associated with malignancy but does not represent true precursor populations of cancer cells. In contrast, small cell change (previously small cell dysplasia) is composed of parenchymal cells of smaller size and larger nuclei (causing nuclear crowding), often containing hepatic progenitor cells and intermediate hepatocytes (transitional forms between progenitor and mature epithelial liver cells). Small cell change foci are considered direct precursors of HCC.3 Macroscopic precancerous lesions are represented by nodules measuring more than 1 mm in diameter, termed dysplastic nodules (DNs), characterized by dysplasia but without definite criteria for malignancy. Many DNs reveal themselves as indistinctly or distinctly nodular lesions ranging from 5 to 10 mm in diameter. They are subclassified into low-grade DNs (LGDNs) and high-grade DNs (HGDNs). LGDNs show only mild dysplasia, and might be confused with large regenerative nodules, corresponding to nodules of ≥5 mm in diameter, larger than surrounding cirrhotic nodules, without atypia, and with no link to neoplasia. In contrast, HGDNs carry a worse prognosis with a four-fold risk of HCC development. They are characterized by clearly higher cell density than that of surrounding tissue, a higher degree of dysplasia, and represent borderline lesions to be differentiated from, but closely resembling well-differentiated HCCs.4 The next step is malignancy, comprising a continuum from in situ HCC and microinvasive HCC to more progressed stages of HCC through ill-defined borderlines in between.3 The transition of HGDN into early malignancy may occur by development of module-in-module (term and concept from Popper5) in which the subnodule often is less differentiated than the original lesion, representing a further step along the pathway of progressing malignancy. Over the years, the entities proposed by the 1995 IWP became more precisely defined by research efforts in different centers. Presently, the diagnosis is based not only on the accurate search of a set of cyto-architectural features (such as cell atypia, cell crowding, trabecular thickness, microacini) but also on a supplement of histochemical (Gomori's silver stain for reticulin fibers) and immunocytochemical stains, including immunostaining of keratins 7 and 19 for evaluating stromal invasion, CD34 immunostaining for visualizing capillarization of sinusoids, and staining for alpha-smooth muscle actin to highlight unpaired arteries that otherwise are difficult to discern. The latter term refers to arteries not accompanied by bile ducts as in portal tracts of normal liver. They reflect neoangiogenesis, increase in number from early to fully malignant nodular lesions, and represent the microscopic counterpart of hypervascularity with washout in the portal/venous phase in contrast-enhanced liver imaging. Of particular importance are histochemical markers of malignant transformation, including heat shock protein 70, glutamine synthetase, and glypican-3.3, 4, 6 Regarding the evaluation of borderline cases, there has been a discrepancy in interpretation between Japanese and Western pathologists. DNs with a vaguely nodular contour, diagnosed as HGDN by Western pathologists, often corresponded to what is recognized as early well-differentiated HCC (eWDHCC) in Japan.7 The reason can be sought in earlier epidemiological studies on the heterogeneous geographical distribution of HCC, mentioning higher incidence in Southeast Asia and Sub-Saharan Africa, and lower incidence in Europe and North America.8 Whereas in the West a lower incidence of HCC may explain a weaker focus of research on its diagnosis and treatment in the past, countries where HCC was previously recognized as a serious threat years ago developed a particularly intense clinical and research activity to combat the disease. In China, population screening for HCC using alpha-fetoprotein testing started in the early 1970s, leading to the clinical detection of HCC before any clinical or even other laboratory abnormalities were evident (the so-called subclinical HCC of Tang).9 Similarly, with considerable stimulation by the late Professor Kunio Okuda, a major effort took place in Japan, starting with follow-up studies in 1965.10 Recognized leaders in the field exchanged data, experience, and proposals, with the result that most of the advances in epidemiology, clinical characterization, staging, prognosis, and treatment of HCC were generated in Japan.11 Special efforts to achieve early diagnosis as a key to successful therapy have enabled surgical resection, nonsurgical treatment, and transplantation of eWDHCC followed by significantly improved survival and even cure.12 This renders identification of this lesion highly important. In the West, a sharper focus on early diagnosis and precancerous lesions was stimulated by a rising incidence of HCC,13 by encouraging results from the East, by accelerating progress in serum markers and in liver imaging, and to a large extent by pathological study of nodular lesions detected in explant livers at the time of liver transplantation. In 2002, Professor M. Kojiro from Kurume University convened a meeting of an International Consensus Group for Hepatocellular Neoplasia (ICGHN) to discuss the diagnosis made by the individual participants on a series of liver nodules circulated before. The main discrepancy was indeed between HGDN and eWDHCC, with the most helpful distinguishing lesion being the occurrence in eWDHCC of tumor cell invasion into the remaining intranodular portal tracts (so-called stromal invasion) (Fig. 1). The latter is originally a Japanese observation,14 which may be difficult to evaluate; it is sometimes as delicate as the recognition of incomplete capsular invasion in follicular carcinoma of the thyroid.15 A small eWDHCC of indistinctly nodular type can be considered the equivalent of carcinoma in situ of other tissues.3 Stromal invasion in nodule of 10 mm diameter in liver with alcoholic cirrhosis. (A) Macroscopy. (B) Histopathology of stromal invasion (hematoxylin-eosin stain). (C) Positive glypican-3 immunostain (courtesy of Prof. Dr. Tania Roskams). After additional meetings in Leuven (Belgium) and Thessaloniki (Greece), the ICGHN has reached sufficient consensus for communicating its proposals to the scientific community by a report in this issue of HEPATOLOGY.16 The state of the art of histologically diagnosing liver nodules has made considerable progress since the first attempt by the IWP in 1995. Of course, as can be expected, several problems remain. The high number of criteria involved in differentiating the various premalignant and early malignant lesions hamper an easy uniform diagnosis. Recognition of stromal invasion requires experience. Furthermore, this lesion may be focal, and hence missing in needle biopsy specimens. The latter are not always easily obtained in ultrasound-guided biopsy techniques, due to their small size and sometimes difficult location of the lesion.17 Nevertheless, biopsy remains indicated in many instances, because insufficient development of unpaired arteries in eWDHCC may hinder neoangiogenesis-based diagnosis of malignancy by noninvasive imaging.17, 18 The need for further study remains the reason for discrepancies in the extent of steatosis in eWDHCC between the East and West, and the search for further and more specific markers for malignancy, vascular invasion, and metastasis.4 Such problems are duely recognized in the paper. Together with the recent subdivision of benign hepatic adenomas according to pathological/molecular criteria, detectable by routine immunohistochemical techniques, especially the recognition of the β-catenin mutated variant with higher risk of malignant degeneration,19 the histopathological diagnosis of early and equivocal hepatic lesions has advanced remarkably. The price paid is an ever-increasing complexity of investigation, necessitated by progress in tumor biology in general, as expressed on the occasion of the centenary celebration of the Pathological Society of Great Britain and Ireland: “In the future it seems likely that a panel of markers will be evaluated in many human cancers as adjuncts to the standard prognostic parameters established by routine morphological assessment. As we move into this new era of rational targeted therapies, meticulous attention to precise tumor classification remains critical for the development of such therapies, and surgical pathology will undoubtedly continue to occupy a central role.”20