Afectaciones ambientales

Based on the assumption that a neoplasm must be composed of a monoclonal cell population, and following the pioneering studies on haematological malignancies, the possibility of assessing the clonality of solid tumours was originally viewed as a

tool for the understanding of carcinogenesis and, in the field of Histopathology, as a potential ancillary method to conventional histological assessment.

Methods based on PCR such as HUMARA and PGK assay have the advantage of using a small amount of DNA, compared to the traditional Southern blotting methods used to assess the integration patterns of HBV, thus offering the possibility of being used on biopsy material. In an article published in 1995 Kawai et al (Kawai et al. 1995) conclude their work stating: "this method (i.e. X- chromosome-linked PGK assay) enables us to analyse the clonality of small materials like biopsy specimens. It may be helpful in certain cases in which the pathological findings are ambiguous, and its use in conjunction with histological appearances would seem imperative in order to avoid errors secondary to inflammatory infiltrates" (referring to the single HCC in their study that showed a polyclonal pattern). As discussed above, even though the clonality assessment by X-chromosome inactivation methods tends to show an increase of monoclonal patterns when MRN and HCC are considered, respectively, the results are not absolute and many identical lesions in terms of size and histological features may show opposite clonal patterns. This may be partly explained by the technical limitations of this method of investigation.

There are limitations intrinsic to the techniques based on X- chromosome inactivation and used to assess clonality (Garcia et al. 1999; Going et al. 2000) as discussed in chapter 1. In particular the patch size question is fundamental for the interpretation of clonality. As X-chromosome inactivation in female individuals occurs at early embryonic stage, groups of cells deriving from division of progenitors with random inactivation of the X-chromosome will show the same X- chromosome inactivation. Ochiai et al (Ochiai et al. 2000) have recently addressed this question. The authors found that randomly sampled areas of background non-cirrhotic liver in patients with haemangioma, cholangiocarcinoma or metastatic adenocarcinoma were monoclonal, and estimated the monoclonal patch size as 1.1±0.3 mm^ (average±S.D.). Samples of liver tissue from a non­ cirrhotic patient with chronic viral hepatitis showed that the presumed monoclonal

patch size was larger than normal liver tissue and was estimated as 3.3±0.9 mm^ The estimated patch size in cirrhotic liver was 31.2± 40.11 mm^. Therefore, like rodent liver, human liver tissue is organized in mosaic patches with differing X- chromosome inactivation patterns that progressively expand during the progression of chronic liver disease.

This may explain the findings of this work and previous studies (Paradis et al. 2000; Aihara et al. 1994; Yeh et al. 2001). Instead of reflecting the biological birth and evolution of neoplasms, these findings may simply indicate the enlargement of pre-existing patches, as part of the regenerative process characteristic of the cirrhotic condition. In particular the size of the RN in this study ranged from 1 to 3 mm diameter, therefore within the size range of the estimated patch size from normal to cirrhotic liver, as estimated by Ochiai et al (Ochiai et al. 2000), see above. On the other hand, the polyclonality of a percentage of nodules, may be explained by the fact that these nodules may have originated from the junction of monoclonal patches (Garcia et al. 1999; Ochiai et al. 2000). Paradis et al (Paradis et al. 2000) have shown that the monoclonal RN are significantly larger than polyclonal ones (mean size of the monoclonal nodules: 3 + 0.1 mm vs mean size of the polyclonal nodules: 2.5 +/- 0.1 mm, p = 0.007), although they are still within the range of the estimated patch size in cirrhotic livers.

The same considerations (and in particular the patch size) for the RN, should be applied to MRN and HCC. In the present study, all "monoclonal" MRN except 1 (RFH patient 3, 15 mm diameter), were above 6 mm diameter, and therefore would not be included in the average size range of the patch size (31.2± 40.11 mm^, see above), as estimated by Ochiai et al (Ochiai et al. 2000), in cirrhotic liver. Only one (6 mm diameter) monoclonal MRN could be comprised within two standard deviations of the mean patch size area in cirrhotic liver as calculated by Ochiai et al. Other studies investigating clonality in cirrhotic livers have been performed on nodules of more than 6 mm diameter (Piao et al. 1997; Paradis et al. 1998) (see table 1.2.1). In the study by Ochiai et al (Ochiai et al. 2000) 58.9%

of MRN (mean area 4.8 ± 6.2 mm^ diameter not given) were monoclonal. Specifically in relation to their estimate of patch size in cirrhotic liver, the authors conclude that most RN showing a monoclonal pattern may have a multiple-cell origin. It is therefore evident that before interpreting monoclonality of MRN in cirrhotic liver, the caveat of the patch size question has to be addressed.

The median size of HCC in the present study was 25 mm, and the smallest HCC was 13 mm diameter, therefore in excess of the patch size area in cirrhotic liver as estimated by Ochiai et al (Ochiai et al. 2000). Thus, the monoclonality of these HCC could not be related to the monoclonal patches. Whether lesions originated within the boundaries of patch maintain the same type of X-inactivation, even when they exceed the patch size limit as a result of proliferative expansion, is speculative. Probably this matter is not relevant for overt HCC, but may help, once again in interpreting the monoclonal pattern of RN.

However there are other possible explanations of monoclonality in HCC. These include increases and reductions of méthylation during progression of malignant tumours; some embryonic tumours such as retinoblastoma show the preferential loss of maternal and paternal alleles; and X-chromosome inactivation might be non-random in some tissues (Garcia et al. 1999).

In the case of tumours of polyclonal origin, clonal evolution may select a dominant clone that will prevail on the others generating a "pseudomonoclonality" (Garcia et al. 1999; Wasan et al. 1997). Some studies have shown that X-inactivation might be non-random, and in particular it has been shown that normal haemopoietic tissue have skewed X-inactivation, possibly favouring the paternal or maternal X- chromosome(Garcia et al. 1999). This, along with the recent appreciation of the relationships between bone marrow and liver regeneration (Alison et al. 2000; Alison et al. 2001; Theise et al. 2000a; Theise et al. 2000b; Petersen et al. 1999), further complicates the issue of clonality in liver tissue. The possible role of hepatic progenitors in hepatocarcinogenesis has been discussed in chapter 1. If hepatic progenitors are of bone marrow origin and are involved in the genesis of

HCC, then it should be taken into account that studies have shown that in normal haemopoietic tissue X-inactivation occurs in a non-random fashion favouring the paternal or the maternal X-chromosome (Garcia et al. 1999). Aihara et al (Aihara et al. 1994) have shown that adjacent monoclonal regenerative nodules inactivate the same allele, and have interpreted this finding with the hypothesis that an expanding monoclonal population of hepatocytes is separated into different nodules by the formation of fibrous septa. An alternative explanation could be that all those monoclonal nodules originate separately either from a single or from multiple extrahepatic progenitors with the same X-chromosome inactivation pattern.

Polyclonality in MRN and HCC (as in the 5 cases described by Piao et al (Piao et al. 1997), may be due to origin at the boundaries of two adjacent patches (Garcia et al. 1999). Other explanations include contamination by non-tumour cells (Kawai et al. 1995), collision of independent tumours, inversion of X chromosome inactivation, transformation in the embryo prior to Lyonization or DNA transfer to bystander cells (Going et al. 2000: Heim et al. 2001; Going et al. 2001).

The patch size question has not been solved. The estimate of the patch size by Ochiai et al (Ochiai et al. 2000) was not a direct observation but was based on the mathematical calculation described by Hutchison (Hutchinson 1973), taking into consideration the proportion of polyclonal samples and the mean microdissected sample size. Moreover, the location of the monoclonal areas in the liver tissue examined, in terms of liver acini or lobules remained unclear.

The limitation of patch size cannot be applied to the studies based on HBV integration pattern such as the studies by Aoki and Robinson (Aoki et al. 1989), Tsuda et al (Tsuda et al. 1988), and Yasui et AI (Yasui et al. 1992). However studies based on virus integration pattern may underestimate the proportion of monoclonal lesions, because they would take into account the proportion of

proliferating hepatocytes without viral integration (Ochiai et al 2000; Aihara et al. 1994).

CHAPTER 6

Assessment of the patch size in