Chromosome 1q Amplification Is Associated with a History of Prior Malignancies Among Patients Newly Diagnosed with Multiple Myeloma

Background: Over the past 20 years, observational data from usual care clinical oncology settings has been leveraged to inform estimates of cancer treatment-associated benefits and risks among patients not treated on clinical trials. Increasing genomic testing to inform treatment decisions in usual care settings now meaningfully augments traditional observational data, positioning it to provide insights beyond clinical care into tumor biology. We studied patients with newly diagnosed multiple myeloma (MM), comparing cytogenetic test patterns according to history of prior malignancy.

Methods: In this retrospective cohort study, we identified 2,380 patients from the COTA real-world database (RWD) who were newly diagnosed with MM in the years 2010-2018. The COTA RWD is a de-identified composite of both abstracted electronic health record and administrative data pertaining to patients receiving their cancer care at one of COTA's clinical oncology practice partners. Among these patients, 1769 (74%) had evidence of MM-associated cytogenetic testing with fluorescent in-situ hybridization (FISH) within the 120 days surrounding their date of diagnosis. The 1,769 patients form the analytic cohort. We compared patients' FISH results for t(4;14), deletion(17p), t(14;16), deletion(13), t(14;20), t(6;14), t(11;14), deletion (1p), and amplification(1q) according to their history of prior malignancy.

Results: Within the cohort, 263 prior malignancies were identified in 241 patients (14%, 241/1,769). Two-hundred and twenty-one patients (92%) had one prior malignancy, 28 (7.9%) had two prior malignancies, and one (<1%) had four prior malignancies. The most common prior malignancies were prostate (n=50), breast (n=19), melanoma (n=14), skin (n=13), and cervix (n=6). Amplification of the long arm of chromosome one (amp(1q)) was noted in 31% of patients (75/241) with a prior malignancy vs. 24% of patients (370/1,528) without (chi2 test p=0.02). Overall 25% of patients had amp(1q). No other translocations, amplifications, deletions were associated with prior cancers. A non-parametric test for trend revealed a strong positive association between patients' malignancy count (range 0-4) and amp1q (p<0.01). MM patients with prior lymphomas and prior melanomas also had high rates of amp(1q), though these were not significantly different from patients without these prior malignancies. In a multivariable logistic regression model that adjusted for patient demographic attributes, other known potentially collinear MM poor prognostic factors (i.e., revised ISS stage, IgA sub-type, lambda light chains) and adjusted standard errors for clustering of patients within treatment settings, a history of prostate cancer remained clinically and statistically significantly positively associated with amp(1q) (OR 2.1, 95% CI: 1.9-2.2) as did history of two or more prior malignancies (OR 2.8, 95% CI: 2.3-3.3). Of note, amp(1q) was positively associated with IgA subtype (OR 1.5, 95% CI: 1.3-1.6) and the presence of lambda subtype (OR 1.3, 95%CI: 1.3-1.4).

Conclusions: Using RWD, we found that newly diagnosed MM patients with histories of prostate cancer and those with two or more prior malignancies were more likely to have amp(1q), a poor prognostic marker in MM. Gains in 1q have previously been identified among patients with prostate and lymphoid cancers, but to our knowledge this is the first study to identify an association with a prior history of cancer, especially prostate cancer, and amp(1q) in MM. This relationship is worth further exploration of whether there is a common pathway associated with for example risk of prostate cancer and amp(1q) in MM. Clinical trials are less likely to answer this question as patients with prior malignancies are often excluded from enrollment. Overall, the results reported suggest that RWD is an efficient and comparatively inexpensive tool to support research in cancer biology through hypothesis generating and testing analyses of linked real-world phenotypic and genotypic data.