Chromosomal profiles of CIN2/3 infected with different hrHPV types were compared in order to determine the potential contribution of different hrHPV types to the heterogeneity in chromosomal profiles as observed previously in CIN2/3 [21
Unsupervised hierarchical clustering analysis revealed considerable heterogeneity between CIN2/3 with respect to their chromosomal aberrations, with one subset having relatively few aberrations (cluster 1) and the other with an increased number of aberrations (cluster 2), similar to our observations in earlier studies [21
]. It should be noted that, even though all lesions were associated with long-term hrHPV infections (≥ 5 years), there was still a subset of lesions with few chromosomal aberrations. This is in line with our previous study comparing lesions with < 5 years versus ≥ 5 years preceding hrHPV infection. In addition to the majority of lesions with short-term infection (< 5 years), also a subset of lesions with long-term HPV infection (≥ 5 year) showed rather few chromosomal aberrations as well. Interestingly, the majority of lesions with HPV16 (87.5%) were in the relatively quiet cluster (cluster 1) and had few aberrations. Within this cluster a subgroup could be recognised that consisted primarily of lesions with HPV16 that had gain of 3q and/or 1. The fact that CIN3 with a short-term HPV16 infection, likely reflecting the fast-progressing HPV16 lesions, also showed few chromosomal aberrations [6
], indicates that the overall detection of fewer aberrations in HPV16 positive lesions is not dependent on duration of HPV16 infection. It should, however, be noted that only women over 30 years of age were included in the study since they were derived from the POBASCAM trial.
The detection of fewer copy number aberrations in HPV16-positive lesions was furthermore corroborated by hrHPV-typing analysis of an independent set of CIN2/3 previously analysed by arrayCGH [21
], which also demonstrated that the majority of CIN2/3 with HPV16 (81.8%) clustered together and showed relatively few aberrations (data not shown).
Related to the clustering results, the average percentage of chromosomal losses in lesions with HPV31 was three-fold higher compared to the HPV16-positive lesions, i.e. 6.7% versus 2.2% (Table ). No conclusion on the distribution over the two clusters of the remaining lesions harbouring other hrHPV types could be drawn due to their low numbers. However, lesions harbouring α5&α7 types had two-fold more losses compared to α9 positive lesions, i.e. 8.5% versus 4.2% (Table ).
The observation that lesions with HPV16 tended to contain fewer aberrations than those with different types may indicate that HPV16 causes faster progression than other types and/or may not need as many chromosomal aberrations for progression to CIN2/3. The idea that lesions with HPV16 may progress faster seems to be supported by data from Vinokurova et al
], who reported a significant difference in the age of cervical cancer patients harbouring different hrHPV types. While the age at CIN3 diagnosis was not significantly different for HPV16-positive women compared to HPV31-positive women, women with an HPV16-positive carcinoma had a median age of 43 years, whereas those with HPV31-containing carcinomas were on average 64 years of age (p
< 0.01). Also, other studies reported hrHPV type-related differences in that women with HPV16-and HPV18-positive carcinomas were younger than those infected with other types [26
Comparison of lesions with HPV16 to HPV31 revealed significant differences in the affected chromosomal regions. In particular, losses at chromosomes 2q, 4p, 4q, 6p, 6q, 8q & 17p and gain of 1p & 1q were significantly more frequent in lesions with HPV31. Ingenuity Pathway Analysis indicated that genes located within significantly more frequently altered chromosomal regions in HPV31-positive lesions were particularly involved in pathways related to the immune response, most of which are related to loss of the locus encoding the major histocompatibility complex molecules (human leukocyte antigen, HLA) at chromosome 6p.
While not significant, lesions with HPV16 had the highest incidence of 3q gain, which was almost two-fold higher compared to HPVnon16
or HPV31. Gain of 3q is one of the most consistent chromosomal aberrations in cervical carcinoma [40
] and has been suggested to predict progression of CIN [47
]. Genes located at 3q, as was recently shown for PIK3CA [48
], may have an important role in malignant transformation. It is tempting to speculate that HPV16-positive lesions with this particular aberration may not require many additive copy number aberrations for progression.
It should be noted that by arrayCGH analysis only copy number aberrations can be detected. Hence, it cannot be excluded that lesions with HPV16 have other, potentially more subtle genomic or (epi)genetic aberrations, such as mutations, loss of heterozygosity or DNA methylation. The fact that HPV16E6E7 have convincingly been demonstrated to induce genetic instability in in vitro
model systems (reviewed by Korzenievski [49
]), would argue for the presence of a genetic instable environment in HPV16-positive CIN3. We anticipate that the aberrations detected with arrayCGH result from their selective growth advantage becoming evident upon selection pressure during multiple cell divisions. This is substantiated by a recent paper by Bester et al
] showing an immediate induction of replication induced DNA damage upon HPV16E6E7 expression, whereas loss of heterozygosity and copy number variations only became evident after 100-250 population doublings.
Immune evasion is important for the persistence of HPV and may be achieved by various mechanisms, including HLA loci being affected by chromosomal loss, loss of heterozygosity, viral integration or mutations, which may result in functional loss [51
]. Indeed expression of HLA class I has been reported to decrease progressively with cervical lesion grade, [59
]. Immune evasion may also occur via direct interaction between viral and host proteins. For HPV16, for instance, direct interaction of E5 with the hydrophobic domain of HLA class I heavy chain was found to prevent its transport to the cell surface [61
], whereas HPV16 E7 can trigger down-regulation of HLA class I
gene transcription [62
]. However, neither the mechanisms nor efficiency of HLA class I down-regulation has been investigated for HPV31 and as such it is not feasible to perform comparisons to HPV16. It may be that the respective viral proteins of other hrHPV types, such as HPV31, are less capable of interfering with antigen presentation, so that persistence in transforming infections becomes more dependent on (epi)genetic aberrations within the HLA locus. Interestingly, lesions with HPV31 also had significantly more losses at the location of the p53
gene on chromosome 17. It is still questionable to what extend p53 activity is affected in these cases. It may be that the E6 protein of HPV16 is more efficient in inactivation of p53 at the posttranslational level than E6 of HPV31 and that for this type additional events are needed to inactivate p53
. Indeed, the study of Ku et al
. reported that in cervical carcinomas the majority of cases with p53
mutations and LOH at 17p occurred in carcinomas harbouring types other than HPV16 [63
]. However, in vitro
experiments do not provide evidence of differences in degradation efficiency of the E6 proteins of these different hrHPV types [64
To the best of our knowledge, there are no other studies in which the relation between specific hrHPV types and chromosomal aberrations in precursor lesions has been examined. So far, only a study on cervical carcinomas demonstrated the presence of type-dependent chromosomal aberrations [65
]. Carcinomas without detectable HPV or containing HPV18 in a single or multiple infection showed a higher incidence of gains at 20q compared to cases containing HPV16 or other types [65
]. Comparison of the frequency of 20q gain in CIN2/3 with HPV16 versus HPV18 was not possible in our data set due to the low numbers of lesions harbouring HPV18. Comparison of the frequency of 20q gain in HPVα9-versus HPVα5&α7-positive lesions, however, indicated an increased frequency in lesions harbouring types from the α5 and α7 species (i.e. 25.8% vs. 36.7%). Gain of 20q was also more frequent in lesions with HPVnon16
or HPV31 in particular, compared to those with HPV16 (29.6%, 35.7% vs. 25.0% respectively).