This study elucidates the impact opportunistic genetic screening of actionable genes has on patients and relatives in a country with free nationwide health care. In 595 pediatric oncology patients that underwent opportunistic genomic screening of 36 medically actionable ACMG genes, 19 secondary findings were reported in genes not associated with cancer or cancer predisposition syndromes; all of which were detected in genes associated with cardiovascular, lipid disorder or connective tissue disorders. The secondary finding rate of 3.2% is higher than previously reported, e.g. by Haer-Wigman et al, in which the detection rate in 1640 healthy individuals was 1.5% in cardiovascular and connective tissue genes.18 This may be due to the smaller sample size in the current study or differences in population frequencies of certain cardiovascular diseases, such as lipid disorders, which are more prevalent in northern Europeans and not reported in the Haer-Wigman paper.19 A recent study of a pediatric cohort from Germany reported a frequency of secondary findings (2,6%) in non-cancer related genes which is similar to the current study.20 Differences in variant interpretation may also play a role: if only pathogenic variants were reported, the rate would have been 1.2%, which is more in line with previous reports.6 A special design of the current study was that the family history was considered during the variant interpretation process, which may have resulted in higher return rate. Knowing that it was possible to confirm the pathogenicity of a variant in a lipid disorder gene with a lipid profile in a patient, to potentially reduce morbidity and mortality with medical treatment, may have outweighed the concern of burdening the patients with unnecessary testing. The fact that all patients had childhood cancer may also have increased the tendency to report the finding, since the likelihood of the patient being exposed to QT-prolonging or cardiotoxic compounds must be assumed to be much higher than in the general population.21 In this paper, only findings in non-cancer genes are reported, which makes the findings less generalizable, regarding the scope of secondary findings as a consequence of opportunistic screening, as many (23/59) of the actionable genes are cancer genes. Also, this was a population with low rates of consanguinity, hence the prevalence of recessively inherited metabolic disorders may be higher in other populations.
In this study, to the uptake was 2.25 pr index patient during an average of 1.6 years of follow up. In seven of these families, cascade testing is still ongoing, why this number is expected to rise. Frey et. al studied uptake of cascade testing and surveillance participation in a cohort of relatives to patients with pathogenic variants in hereditary cancer syndromes with an observation time of two years and similarly found that an average of 1.9 relative pr index case participated in that time.22 Manchester Genomic Medicine found that 3.05 relatives were tested per family with an actionable cancer predisposing variant identified over a follow up period of 30 years.23 While these numbers are not directly comparable, they point to the fact that in a setting of public health care and good access to genetic counseling, the effect of finding an actionable variant, whether as a primary or an secondary finding, both in terms of potential health interventions and health service provision, extends beyond the effect for the index case.
In the probands, uptake to genetic counseling or clinical testing for the secondary finding was high (84%, 16/19). We are aware of four relatives that actively declined genetic testing when learning about the option. It is a weakness of our design that we cannot distinguish between relatives that have actively decided not to get tested and relatives that simply have not yet been tested. In general, uptake of genetic counseling in at risk relatives varies significantly in different studies and between diagnosis. In a recent review of genetic testing in relatives from families with hereditary cancer, uptake was found to be 48% (95% CI 38–58) overall.24 In a similar metaanalysis of relatives in families with hypertrophic cardiomyopathy uptake varied from 37–84%.25 It appears that uptake was high in our cohort. We propose that this is due to the direct contact with the families and the nature of the public health care system in Denmark. Families that recently have experienced pediatric cancer may be less worried about a hypothetical risk of cardiovascular disease. Conversely, the families’ behavior and responses regarding the identification of a secondary finding may be atypical due a challenging situation with either on-going cancer treatment, terminal disease, or follow-up in survivors. Further studies of the patient perspective may elucidate this in the future.
Medical actionability is a key argument for the application of OGS and the return of secondary findings. In this study, medical interventions such as regular follow up, dietary or pharmacological treatment were initiated in 61% of the probands in which the secondary finding was disclosed, as well as in 16 relatives (42% relatives). Interestingly, neither patients nor relatives underwent invasive procedures or the implantation of medical devices. This of course would be expected to be different if cancer predisposition genes such as BRCA1 or BRCA2 were included in the return of secondary findings, as in a recent paper on actionable genomic findings in the BabySeq project, where the return of a genomic finding in children led to risk reducing surgeries in parents.26 In several families there was a family history of cardiovascular disease, but not such that the family had been diagnosed with a specific hereditary disease. This finding emphasizes the fact that many secondary findings can be expected to be disclosed to patients or families not previously aware that they were at risk of an inherited disease. Counseling in these patients and families is also complicated by paucity of penetrance estimates for many diseases in patients without relevant family history.27 This dilemma was discussed specifically in two families where a pathogenic variant was disclosed in a gene associated with reduced penetrance (SCN5A and PKP2). After clinical evaluation and dialogue with the family (shared decision making), further cascade testing was not performed.
Likewise, it is important to highlight that no treatment or follow up was planned in 29 individuals (7 patients, 22 relatives) after they had been evaluated by a clinical geneticist or by another specialist. This was both due to a negative predictive genetic testing result (14 relatives), changes in variant classification (3 probands, 4 relatives), variant interpretation related to the disease phenotype (2 probands) or clinical evaluation (1 proband, 4 relatives), or wishes from the family (Shared decision making in one proband, wished no further follow up). Variant interpretation is a common challenge in the field of clinical genetics, both in diagnostic testing as well as in a setting of OGS as reported by Hart et al., where 14 variants of the 76 reported variants (18%) were reclassified as a variant of unknown significance after evaluation of new evidence.7 Further research in the related psychological and emotional impact in these individuals will be interesting.
While the ACMG position statement provides guidance for actionability in a setting of OGS, certain genetic variants may be more or less “actionable” in different populations or patient cohorts. For example, patients with Charcot-Marie-Tooth disease due to duplication of the PMP22 gene are at higher risk of permanent neurological complications after treatment with certain antineoplastic agents.28 Knowledge of such a predisposition could potentially be useful when planning treatment in order to minimize side effects in cancer patients. Recently, the list of medically actionable genes was expanded to encompass 81 genes, and could potentially grow after each version, as more knowledge of genotype-phenotype relationships are understood and novel medical therapies become available.3,4,29 As reported by Johnson et al, the returnable variant frequency rate increased by 22% when 73 actionable genes from the third version of the ACMG actionable list were analyzed compared to the 59 genes from the second version.30 If implemented, this can be expected to increase the burden on both variant interpretation laboratories and genetic counseling resources as well as impact even more patients and their families.