HLA-A*02 and its prognostic traits in cancer : an immunological biomarker as a tool towards individualised cancer therapy

Abstract

This thesis is based on the findings that HLA-A*02, a common allele in the Swedish population, has a decreasing frequency in the European continent in relation to geographic latitude. Furthermore, mortality of ovarian cancer correlates to the HLA-A*02 frequency and geographic latitude.

The aims of the thesis were to study the HLA haplotypes and overrepresentation of HLA-A*02 in ovarian cancer patients, to determine the prognostic traits of HLA-A*02 in combination with MHC class I expression, as well as to analyse the role of HLA-A*02 in colon cancer patients together with MHC class I expression and CD8+ lymphocyte infiltration.

In the first paper, HLA-A*02 is determined in 97 ovarian cancer patients and correlated with overall survival. HLA-A*02 was analysed from FFPE tumour samples. Patients with serous ovarian cancer and HLA-A*02 positivity had a poorer survival compared to those who were HLA-A*02 negative.

In the second paper, blood samples were collected from 32 patients with advanced epithelial ovarian cancer (EOC) to determine HLA-A, -B, -Cw and -DRB1. The results showed an unusual representation of HLA alleles in EOC compared to healthy Swedish bone-marrow donors where HLA-A*02 was overrepresented, also HLA-A*02 homozygotes were two-fold higher among EOC patients. The frequency of HLA-A*01 was also increased whereas HLA-A*03 was decreased. Combinations of A*02 with B*05, B*15, DR1*03, DRB1*04, Cw*3 and DRB1*03 were also significantly increased, but only when considered as single, non-corrected analysis. These results indicated that the prognostic trait was restricted to HLA-A*02.

In the third paper, blood samples and FFPE tumour samples were collected from 162 EOC patients and analysed for HLA-A*02 genotype as well as MHC class I and β2 microglobulin expression. Patients were with advanced stage, serous histology and HLA-A*02 positivity were grouped and named “worse prognosis group”. In this group there was an increased downregulation of MHC class I, which also correlated to a poorer prognosis. However the level of MHC expression was independent of prognosis when excluded from the worse prognosis group, suggesting that the prognostic trait of HLA-A*02 was not dependent on MHC class I expression.

In the fourth paper, FFPE tumour samples were collected from 520 patients with colon cancer randomised to surgery and adjuvant chemotherapy or surgery alone. Tumour slides were analysed for HLA-A*02 genotype, MHC class I expression, HLA-G expression and CD8+ lymphocyte infiltration. Expression of HLA-G was a negative prognostic marker for the male patients. Also a high infiltration of CD8+ lymphocytes was important in the male patients, where a high frequency of infiltration correlated with a good prognosis. The HLA-A*02 genotype was, on the other hand, a superior negative prognostic marker for the female patients compared to CD8+ lymphocyte infiltration.

In the fifth paper, blood samples and FFPE tumour samples were collected from 16 EOC patients. Blood samples were HLA typed at the clinical immunology laboratory. DNA was extracted from the FFPE samples and primers were designed to amplify exons 2, 3 and 4 of HLA-A. The amplification product was then sequenced and the sequence was aligned against all known HLAalleles. Complete HLA-A alleles could be assigned to 14/16 patients, in one case only one allele could be assigned and in one case no alleles could be assigned. The conclusion is thus that HLAA typing can be performed in fragmented FFPE-derived DNA using this approach.