Defining New Colorectal Cancer Syndromes in a Population-based Cohort of the Disease

Discussion

Known cancer syndromes often involve an increased risk for a whole spectrum of tumours, such as CRC, endometrial, gastric, renal pelvis and other tumours in LS and breast cancer, leukemia, sarcoma, as well as brain tumours in Li-Fraumeni syndrome (12, 13). Also, for the BRCA1, BRCA2 and APC genes and, in fact, almost all known cancer genes, a typical spectrum of different cancers is associated with each gene involved in the syndrome (14-16).

When CRCs cluster in families where none of the known syndromes are prevalent, other tumours are also frequently seen. To find out if this was significant, we used a cohort of consecutive CRC cases and their family history of cancer among close relatives for the study. After comparing the family history it was clear that several tumours were more prevalent in the families with more than one CRC case. It is difficult to determine whether there was only one cancer family syndrome with a differently increased risk for many cancers, or several different syndromes – including several different tumour spectra. We tested tumour types separately and found a positive statistically significant association for urinary bladder, prostate and gastric cancer, melanoma, leukemia/lymphoma groups and gynecological cancer. Compared to the strong association with even rare tumours such as urinary bladder and gastric cancer, the association with breast cancer is weaker and more uncertain. It is at least likely that specificity for one or several CRC syndromes exists, but without knowledge of the underlying genetic contribution, it is impossible to say what tumours are associated with each syndrome.

One limitation of our study is that not all the diagnoses among family members were verified through medical records. However, all abdominal malignancies with an unclear diagnosis were verified in order to confirm or exclude CRC. Other diagnoses were coded as reported from the index patient, if stated in detail and claiming good knowledge. Weak remembrance or uncertainty resulted in our not coding a cancer diagnosis. Some malignancies were considered more uncertain than others. So, for example, gynecological malignancies and hematological malignancies were often stated as such and we rarely specified in detail why we chose to use these terms for all cases reported, regardless of how specifically the diagnoses were expressed. Another limitation is that we did not adjust for differences in family size. It is possible that some families reported more cancers because of larger family size; thus, large families with at least two cases of CRC should show more cancers. However, the prevalence of two close relatives with CRC is associated with an increased risk of CRC for other close relatives, but was not known until now to be associated with other cancers (4).

We used family history and tumour testing to select cases for genetic testing and diagnosis of Lynch syndrome. Only about 1.2% of the patients in Sweden should have LS, judging from a previous study (17). Using the strategy in the present study excluded 1% of the cases with Lynch syndrome based on molecular genetic testing for Lynch syndrome. Even if our strategy might have missed a few cases, such a small proportion is not likely to influence our results. Considering the results and the suggested syndromic tumours, only gastric cancer is clearly associated with LS (12); however, it is rarely seen in Swedish families. Melanoma has not been reported to be overrepresented in LS, and prostate cancer has been reported to be more common in LS-gene carriers than among the general population (18), although this might relate to the selection of families (19). Gynecological cancer constitutes typical tumours of the LS, but endometrial cancer was associated with CRC in both LS and non-LS families who have undergone genetic testing at our clinic at Karolinska Institute, Sweden.

An effort was made to find more evidence in support of the suggested new syndrome(s). Predisposed CRC genes are typically morphogens, thus CRC tumours will demonstrate different morphology depending on the underlying genetic contribution (20). Morphological data was available from half of the patient material (11). Tumour morphology in the index patients was used for testing the hypothesis that tumours in five different syndromes might show different and typical phenotypic characteristics to support different underlying genetic etiology. Gynecological malignancies and hematological malignancies were excluded in the subsequent morphology study. Both groups included tumours not specified in detail and were therefore possibly heterogenous and more uncertain. Breast cancer was also excluded due to its weak association to CRC (p=0.023), even though it is a common cancer. We found statistically significant associations for four of five hypothetical syndromes tested: cancer families, CRC-prostate cancer families, CRC-melanoma families and CRC-gastric cancer families. Since this is the first study of a detailed morphology associated with new CRC cancer syndromes, we thought it was relevant to show all possible results and chose to not correct for multiple testing; therefore, false positives cannot be ruled out from some of our results. The findings included few of the 11 tested tumour characteristics, which does not provide strong evidence for any specific syndrome. Nevertheless, the results still support a different underlying genetic cause of putative cancer syndromes.

Data form this study suggest that there are one or more different colorectal cancer syndromes (Lynch syndrome excluded) involving gastric cancer, prostate cancer, urinary bladder cancer, melanomas, leukemia/lymphosas, gynecolocical cancers and/or breast cancer.