Abstract
Background
Myotonic dystrophy type 1 (DM1) is an inherited trinucleotide repeat disorder in which specific cancers have been implicated as part of the disease phenotype. This study aimed to assess whether cancer risk in DM1 patients is modified by disease severity.
Methods
Using the United Kingdom Clinical Practice Research Datalink (primary care electronic medical records), we identified a cohort of 927 DM1 and a matched cohort of 13 085 DM1-free individuals between January 1, 1988 and February 29, 2016. We used Cox regression models to calculate the hazard ratios (HRs) and 95% confidence intervals (CIs) of organ-specific cancer risks. Analyses were stratified by age at DM1 diagnosis as a surrogate for disease severity. Statistical tests were two-sided.
Results
Patients with classic DM1 (age at diagnosis: 11–40 years) were at elevated risk of cancer overall (HR = 1.81; 95% CI = 1.12 to 2.93); cancers of the thyroid (HR = 15.93; 95% CI = 2.45 to 103.64), uterus (HR = 26.76; 95% CI = 2.32 to 309.26), and cutaneous melanoma (HR = 5.98; 95% CI = 1.24 to 28.79) accounted for the excess. In late-onset DM1 patients (age at diagnosis >40 years), a reduced overall cancer risk was observed (HR = 0.53; 95% CI = 0.32 to 0.85), possibly driven by the deficit in hematological malignancies (DM1 = 0 cases, DM1-free = 54 cases; P = .02). The difference between the observed HR for classic and late-onset DM1 was statistically significant (P < .001).
Conclusions
The observed difference in relative cancer risk between classic and late-onset DM1 patients compared with their DM1-free counterparts provides the first evidence that disease severity modifies DM1-related cancer susceptibility. This novel finding may guide clinical management and scientific investigations for the underlying molecular mechanisms in DM-related carcinogenesis.
Myotonic dystrophy type I (dystrophia myotonica I; DM1; Steinert’s disease) is an autosomal-dominant, neuromuscular disorder characterized by progressive muscle wasting, myotonia, and multisystem involvement (1,2). It is caused by an unstable cytosine-thymine-guanine (CTG) trinucleotide repeat expansion in the 3’ untranslated region of the dystrophia myotonica protein kinase (DMPK) gene on chromosome 19 (3,4). Although rare, DM1 is one of the most common forms of adult-onset muscular dystrophy (5). The spectrum of DM1 severity varies widely, with a notable inverse correlation with age at onset, ranging from fatal outcomes in some congenital presentations to mild symptoms in late-onset disease (5).
Recently, five epidemiological studies have documented elevated cancer risk in DM (6–10), making it the first inherited nucleotide repeat disorder found to increase cancer susceptibility. Although most of the studies consistently reported elevated risks of cancers arising in the uterus and thyroid gland, inconsistencies related to other sites including the brain, ovary, colon, testis, and lung were observed. A recent study found DM1 patients to be at high risk of skin cancers, specifically melanoma and basal cell carcinoma (11). The molecular mechanism underlying DM1-related carcinogenesis remains unclear; hypotheses include RNA spliceopathy possibly resulting in modified expression of downstream oncogenes or tumor suppressor genes, altered protein-coding mechanisms, and/or upregulation of β-catenin (a protein wherein mutations are associated with various cancers) (12–15). A possible explanation for the conflicting results of site-specific cancer risk in DM1 patients is that cancer risk may be modified by disease severity given the variable frequency of symptoms across mild, classic, and severe disease (15,16). Therefore, we used primary care medical records from the United Kingdom to evaluate the risk of cancer in the full clinical spectrum of DM1 patients, compared with a matched cohort of DM1-free individuals.
Methods
Data Sources
Established in 1987, the Clinical Practice Research Datalink (CPRD) is one of the largest electronic repositories of anonymized primary care medical records (17). Within CPRD, data are recorded using “Read” codes, a hierarchical clinical classification system used by the UK National Health Services (NHS). Linkage to practice-level Indices of Multiple Deprivation (IMD; a socioeconomic status proxy measure) is available for all CPRD practices. Additional patient-level linkable data sources, including inpatient hospitalization records (Hospital Episodes Statistics [HES]; April 1997–February 2016) and mortality data (Office of National Statistics [ONS]; January 1998–March 2016), are available for about 57% of the CPRD practices (17). Clinical diagnoses in HES and ONS are recorded using the International Classification of Diseases, 10thRevision (ICD-10) coding framework.
This study was approved by the CPRD Independent Scientific Advisory Committee (ISAC; Protocol # 16_005RA2; see Notes for copyrights) and exempted from full institutional review board review by the National Institutes of Health Office of Human Subject Research.
Study Population
Details regarding study design and patient selection are available elsewhere (11). Briefly, we identified all patients with a DM1 diagnostic code (CPRD Read codes: “F392011: Steinert’s disease”, and “F392000: Dystrophia myotonica [Steinert’s disease]”) within CPRD’s October 2016 data release (n = 1061). For each DM1 patient, we randomly selected up to 20 DM1-free individuals from the pool of patients who received their care in the same clinic, never had a record of DM1 in CPRD, and were alive and registered at the index date of their respective DM1 patient (n = 15 130). Index date was defined as the later date of first DM1 record (a proxy for age at DM1 diagnosis) or registration with the clinic if the patient’s first DM1 record was before enrollment. DM1 patients and the comparison cohort were matched on year of birth (± 2 years), sex, clinic, and clinic registration year (± 1 year).
For the current analysis, we excluded 134 DM1 and 2045 DM1-free patients who either 1) had cancer prior to the start of follow-up; 2) had a code for “history of cancer” as their first cancer record or within six months of their first cancer record; 3) had less than six months of follow-up time; and 4) had no matches after other exclusions; or 5) were classified as DM1-free in CPRD but had a myotonic dystrophy hospitalization record (ICD-10 code G71.1). The final analysis included 927 DM1-affected and 13 085 DM1-free individuals (Figure 1).
Figure 1.
Flowchart of the study population. *Among those excluded were 383 individuals free of myotonic dystrophy type 1 (DM1-free) and 36 individuals affected by DM1 who had a record of cancer; however, some of these cancers (DM1= 8; DM1-free = 125) occurred after start of follow-up. † H/O = history of; ICD = International Classification of Diseases, 10th Revision.
Cancer Ascertainment
First-incident cancer was defined as the earliest recorded cancer event (excluding non-melanoma skin cancers) identified from CPRD for patients not eligible for linkage (n = 5777), or CPRD or HES records for those eligible for linkage (n = 8235). Individuals who did not have a record of cancer in CPRD or HES but whose primary cause of death was cancer, as reported by ONS (n = 14/749; 1.9%), were considered as cancer cases, with their date of death as the cancer diagnosis date. From CPRD, a comprehensive list of cancer-related Read codes (available upon request) was used to identify all cancer events. For linkage-eligible individuals, ICD-10 codes C00 to C96 (except C44 and C4A, for non-melanoma skin cancer) were used to identify cancer events from HES and ONS. If the first cancer record was a secondary, metastatic, or unspecified cancer, the patient’s records (in CPRD and HES, when applicable) were reviewed to determine the primary cancer site.
Statistical Analysis
We used Cox proportional hazards regression models to compare the risks of developing cancer (overall and by anatomic site) for DM1 patients and their matched cohort. The proportional hazards assumption was assessed using Schoenfeld residuals; no violations were observed. Age was the time scale used for all analyses. Follow-up started at the latest of age-at-DM1-diagnosis/DM1-free selection, practice registration, or study start date (January 1, 1988; after the start of CPRD). Patients who were eligible for linkage to HES were followed to the earliest of age-at-first-cancer-record, death, or end of HES coverage (February 29, 2016). Dates of death were ascertained from ONS for patients with linkable records, and from CPRD otherwise. Patients who were not eligible for linkage were censored at the earliest of age at death, transfer out of the CPRD practice, last data collection in CPRD, or end of study (February 29, 2016). To accommodate the study design, baseline hazards were stratified on the matched sets, and late entry into the cohort was accounted for in the PROC PHREG procedure in SAS (18). Models were adjusted for the average number of primary care visits per year, calculated as the total number of clinical visits after start of follow-up to one year before cancer or censoring date, divided by the number of follow-up years. To adjust for cancer ascertainment from multiple databases for linked subjects, models were further adjusted for linkage status as a time-varying covariate. When cancers appeared in one subgroup only, we used the Fisher exact test to examine intergroup differences.
Our main analysis was run separately in strata defined by age-at-first-DM1-diagnosis (0–10 years = congenital/childhood; 11–40 years = classic; and >40 years = late onset), a known proxy for disease severity (5). We further stratified the analysis by sex. Heterogeneity of estimates across strata was tested using a Wald test.
Finally, we conducted various sensitivity analyses to evaluate the robustness of our findings. First, we used several more-stringent definitions of DM1 to optimize the validity of the diagnoses, and restricted the analysis to 1) DM1 patients who had a DM1 record in at least two data sources and their matched cohort (DM1 = 376; DM1-free = 5541); 2) DM1 patients diagnosed after their clinic’s “up-to-standard” date (a CPRD practice-level recording quality metric) (17) and their matched cohort (DM1 = 454; DM1-free = 5936); 3) DM1 patients diagnosed in 1995 or later (after DM1 gene discovery in 1992 and subsequent implementation of DM1 genetic testing in the United Kingdom) (19) and their matched cohort (DM1 = 530; DM1-free = 6988). Second, we restricted the analysis to patients who were eligible for linkage to the HES database (DM1 = 558; DM1-free = 7426). Third, we conducted a sensitivity analysis ending follow-up at a maximum of eight years for all patients to assess if differential follow-up time for the DM1 and DM1-free cohorts affected our results. Fourth, we conducted a subgroup analysis in patients diagnosed with DM1 on or after their registration with their current CPRD clinical practice and their matched cohort. These patients had prospectively recorded DM1 and cancer diagnoses, and thus provide more certainty about the exact date of first diagnosis. In this subgroup, we first repeated the analysis using the previously mentioned follow-up timeline in 494 DM1 and 6449 DM1-free individuals. In a second analysis, we started follow-up at clinic registration date (DM1 = 350; DM1-free = 5127), to evaluate if cancer exclusions differentially affected our results in patients with late-onset DM1. Finally, we repeated the analysis excluding cancers identified from ONS only.
Two-sided tests were used with statistical significance defined as P less than .05. Analyses were conducted using SAS version 9.4 (SAS Institute, Cary, NC) and R version 3.3.1.
Results
Of the 927 DM1 patients, 132 (14.2%) were diagnosed between ages 0 and 10 years, 500 (53.9%) between 11 and 40 years, and 295 (31.8%) after age 40 years, with an overall mean age-at-DM1-diagnosis of 31.2 years (SD = 17.8). There were more females than males in DM1 patients with congenital/childhood (56.1%) and classic (53.2%) disease, whereas patients with late-onset DM1 were more frequently males (55.3%). Overall, median follow-up time was similar in the DM1 (8.8 years) and DM1-free (8.7 years) cohorts. As expected, DM1 patients visited their primary care clinics more often than their DM1-free matched cohort (mean primary care visits per year = 10.3 [SD = 11.0] vs 5.0 [SD = 8.3], respectively), an observation that remained consistent after stratification by disease severity. Table 1 summarizes baseline demographic and clinical characteristics of the study population, stratified by categories of disease severity.
Table 1.
Characteristics of the 927 individuals affected by myotonic dystrophy type 1 (DM1) and the 13 085 DM1-free individuals selected for this study, stratified by age at DM1 diagnosis
| Characteristics | Congenital/childhood DM1 | Classic DM1 | Late-onset DM1 | |||
|---|---|---|---|---|---|---|
| (0–10 years) |
(11–40 years) |
(>40 years) |
||||
| DM1-free | DM1 | DM1-free | DM1 | DM1-free | DM1 | |
| (n = 2203) | (n = 132) | (n = 7538) | (n = 500) | (n = 3344) | (n = 295) | |
| Age at DM1 diagnosis, mean (SD), y | – | 2.8 (3.3) | – | 26.6 (7.8) | – | 51.7 (8.4) |
| Age at start of follow-up, mean (SD), y | 16.1 (13.7) | 16.2 (13.4) | 31.6 (10.0) | 32.0 (10.2) | 53.0 (8.3) | 53.9 (8.9) |
| Sex, No. (%) | ||||||
| Male | 967 (43.9%) | 58 (43.9%) | 3633 (48.2%) | 234 (46.8%) | 1874 (56.0%) | 163 (55.3%) |
| Female | 1236 (56.1%) | 74 (56.1%) | 3905 (51.8%) | 266 (53.2%) | 1470 (44.0%) | 132 (45.7%) |
| Cancer | ||||||
| Yes, No. (%) | 18 (0.8%) | 0 (0.0%) | 221 (2.9%) | 21 (4.2%) | 469 (14.0%) | 20 (6.8%) |
| Mean age at diagnosis (SD), y | 33.1 (17.5) | – | 52.7 (11.2) | 47.5 (8.9) | 65.7 (9.5) | 65.1 (13.0) |
| Death during follow-up | ||||||
| Yes, No. (%) | 22 (1.0%) | 13 (9.8%) | 230 (3.1%) | 109 (21.8%) | 587 (17.6%) | 123 (41.7%) |
| Mean age at death (SD), y | 37.0 (21.5) | 37.7 (15.7) | 50.8 (11.8) | 48.6 (10.3) | 68.8 (11.4) | 64.0 (8.8) |
| Country, No. (%) | ||||||
| England | 1786 (81.1%) | 104 (78.8%) | 6385 (84.7%) | 410 (82.0%) | 2653 (79.3%) | 233 (79.0%) |
| Other | 417 (18.9%) | 28 (21.2%) | 1153 (15.3%) | 90 (18.0%) | 691 (20.7%) | 62 (21.0%) |
| IMD (practice), No. (%) | ||||||
| 1 (least deprived) | 225 (10.2%) | 13 (9.8%) | 1159 (15.4%) | 74 (14.8%) | 454 (13.6%) | 39 (13.2%) |
| 2 | 316 (14.3%) | 19 (14.4%) | 1265 (16.8%) | 88 (17.6%) | 617 (18.5%) | 59 (20.0%) |
| 3 | 474 (21.5%) | 30 (22.7%) | 1419 (18.8%) | 91 (18.2%) | 532 (15.9%) | 43 (14.6%) |
| 4 | 615 (27.9%) | 38 (28.8%) | 1603 (21.3%) | 111 (22.2%) | 552 (16.5%) | 46 (15.6%) |
| 5 (most deprived) | 573 (26.0%) | 32 (24.2%) | 2092 (27.8%) | 136 (27.2%) | 1189 (35.6%) | 108 (36.6%) |
| Baseline smoking status*, No. (%) | ||||||
| Ever | 488 (22.2%) | 22 (16.7%) | 2856 (37.9%) | 177 (35.4%) | 1448 (43.3%) | 132 (44.7%) |
| Never | 358 (16.3%) | 32 (24.2%) | 1697 (22.5%) | 143 (28.6%) | 552 (16.5%) | 66 (22.4%) |
| Missing | 1357 (61.6%) | 78 (59.1%) | 2985 (39.6%) | 180 (36.0%) | 1344 (40.2%) | 97 (32.9%) |
| Baseline alcohol consumption*, No. (%) | ||||||
| Ever | 449 (20.4%) | 28 (21.2%) | 2917 (38.7%) | 191 (38.2%) | 1327 (39.7%) | 116 (39.3%) |
| Never | 202 (9.2%) | 16 (12.1%) | 632 (8.4%) | 49 (9.8%) | 227 (6.8%) | 35 (11.9%) |
| Missing | 1552 (70.4%) | 88 (66.7%) | 3989 (52.9%) | 260 (52.0%) | 1790 (53.5%) | 144 (48.8%) |
| Baseline BMI*, No. (%) | ||||||
| Underweight/normal (<25) | 540 (24.5%) | 40 (30.3%) | 2073 (27.5%) | 163 (32.6%) | 665 (19.9%) | 77 (26.1%) |
| Overweight/obese (≥25) | 260 (11.8%) | 16 (12.1%) | 1645 (21.8%) | 102 (20.4%) | 1047 (31.3%) | 89 (30.2%) |
| Missing | 1403 (63.7%) | 76 (57.6%) | 3820 (50.7%) | 235 (47.0%) | 1632 (48.8%) | 129 (43.7%) |
| Avg PCP visits per year, mean (SD) | 4.6 (8.0) | 11.3 (15.3) | 4.5 (8.5) | 9.6 (9.8) | 6.2 (8.0) | 11.0 (10.6) |
| Follow-up time, median (range), y | 8.7 (0.5–28) | 10.0 (0.5–28) | 8.6 (0.5–28) | 9.3 (0.5–28) | 9.0 (0.5–28) | 7.6 (0.5–27) |
| Total follow-up time, person-years | 23 064.1 | 1562.4 | 74 912.7 | 5300.6 | 33 164.7 | 2549.9 |
For these variables, records closest to the start of follow-up (± 2 years) were used. Avg = average; BMI = body mass index; IMD = index of multiple deprivation; PCP = primary care provider.
During 140554 person-years of follow-up, 41 (4.4%) DM1 patients developed cancer, compared with 708 (5.4%) DM1-free individuals (median age at cancer= 54.0 years vs 61.7 years, respectively). Zero cancers were observed in the subset of patients with congenital/childhood DM1 vs 0.8% in matched DM1-free subjects (Fisher exact P value = .62). Compared with matched DM1-free individuals, patients with classic DM1 were at an elevated risk of cancer overall (HR = 1.81; 95% CI = 1.12 to 2.93, P = .02); cancers of the thyroid (HR = 15.93; 95% CI = 2.45 to 103.64, P = .004), uterus (HR = 26.76; 95 % CI = 2.32 to 309.26, P = .009), and cutaneous melanoma (HR = 5.98; 95 % CI = 1.24 to 28.79, P = .03) accounted for the excess. Elevated risks of cancers in the ovary, brain, pancreas, kidney, and colorectum were also observed, but did not reach statistical significance (Table 2). In late-onset DM1 patients, a reduced overall cancer risk was observed (HR = 0.53; 95% CI = 0.32 to 0.85, P = .009), the deficit likely driven by hematological malignancies (DM1 = 0 cases, DM1-free = 54 cases; P = .02), and possibly female breast cancers (HR = 0.23, 95% CI = 0.03 to 1.74, P = .15), cancers of the lung (HR = 0.40, 95% CI = 0.12 to 1.36, P = .14), and prostate (DM1 = 0 cases vs DM1-free = 27 cases; P = .16) (Table 2). The difference between the observed HR of all cancers combined in patients with classic and late-onset DM1 was statistically significant (P < .001). Stratification by sex within categories of disease severity showed no statistically significant sex difference in overall cancer risk in classic DM1 (HR = 2.03, 95% CI = 1.09 to 3.77 in females vs HR = 1.48, 95% CI = 0.68 to 3.23 in males, Pheterogeneity = .28), or late-onset disease (HR = 0.45, 95% CI = 0.22 to 0.90 in females vs HR = 0.62, 95% CI = 0.32 to 1.21 in males, Pheterogeneity = .73).
Table 2.
Adjusted hazard ratios and 95% confidence intervals of cancer risk comparing individuals affected by and free of myotonic dystrophy type 1 (DM1) selected for this study, stratified by disease severity
| Classic DM1(Age at diagnosis 11-40 y) | Late-onset DM1(Age at diagnosis > 40 y) | |||||
|---|---|---|---|---|---|---|
| DM1=500; DM1-free=7538 |
DM1=295; DM1-free=3344 |
|||||
| Site* | Cancers† | HR‡ (95% CI) | P | Cancers† | HR‡ (95% CI) | P |
| All | 242 | 1.81 (1.12 to 2.93) | .02 | 489 | 0.53 (0.32 to 0.85) | .009 |
| Thyroid | 6 | 15.93 (2.45 to 103.64) | .004 | — | — | — |
| Ovary or uterus§ | 10 | 14.88 (2.14 to 103.67) | .006 | 25 | 0.32 (0.04 to 2.62) | .29 |
| Melanoma | 14 | 5.98 (1.24 to 28.79) | .03 | 9 | 1.14 (0.09 to 14.13) | .92 |
| Brain | 8 | 4.99 (0.46 to 53.78) | .19 | 13 | 2.03 (0.23 to 17.68) | .52 |
| Kidney | 11 | 2.26 (0.22 to 23.54) | .50 | 8 | 2.13 (0.21 to 21.81) | .53 |
| Pancreas | 7 | 2.96 (0.30 to 29.38) | .36 | 17 | 1.81 (0.21 to 15.23) | .59 |
| Colorectum | 17 | 1.82 (0.32 to 10.31) | .50 | 61 | 1.12 (0.37 to 3.45) | .84 |
| Esophagus | 8 | — | 1.00‖ | 26 | 1.57 (0.39 to 6.41) | .53 |
| Lung | 33 | 1.21 (0.14 to 10.15) | .86 | 70 | 0.40 (0.12 to 1.36) | .14 |
| Female breast | 48 | 0.52 (0.12 to 2.24) | .38 | 44 | 0.23 (0.03 to 1.74) | .15 |
Only reported sites with a minimum of two cancer patients in each cohort, overall. The analysis of all sites combined included all cancers (not restricted to sites reported here). HR=hazard ratio; CI=confidence interval; — = No events in at least one group.
Reported as the total number of cancers in DM1 and DM1-free patients to adhere to Clinical Practice Research Datalink policy, which prohibits the reporting of cells with fewer than five events.
Baseline hazards stratified on matched sets, model adjusted for average number of primary care visits per year, and linkage status except when indicated. Reference group is DM1-free cohort.
Site-specific hazard ratios for uterine and ovarian cancers in classic DM1 were 26.76 (95%CI=2.32 to 309.26, P=.009) and 6.09 (95%CI=0.48 to 77.75, P=.16), respectively.
P value (two-sided) obtained from the Fisher exact test.
In the sensitivity analysis restricted to DM1 patients with DM1 diagnoses recorded in multiple data sources (Table 3), results for the classic group were consistent with the main analysis (HR for all cancers combined = 1.89, 95% CI = 1.01 to 3.52, P = .05). In patients with late-onset DM1 the HR was slightly higher than observed in the main analysis (HR for all cancers combined = 0.84, 95% CI = 0.42 to 1.67, P = .61). Results were similar in patients with late-onset DM1 when restricted to patients with their first DM1 record at or after clinic registration, with follow-up starting from age at clinic registration (HR = 0.76, 95% CI = 0.38 to 1.55). Results from all other sensitivity analyses were consistent with our main findings (Table 3).
Table 3.
Sensitivity analyses of cancer risk estimates comparing individuals affected by myotonic dystrophy type 1 (DM1) with DM1-free individuals selected for this study, stratified by disease severity
| Sensitivity analysis* | Classic DM1(Age at diagnosis 11-40 y) |
Late-onset DM1(Age at diagnosis > 40 y) |
||||
|---|---|---|---|---|---|---|
| Cancers† | HR‡ (95% CI) | P | Cancers† | HR‡ (95% CI) | P | |
| 1 | 133 | 1.89 (1.01 to 3.52) | .05 | 183 | 0.84 (0.42 to 1.67) | .61 |
| 2 | 49 | 1.77 (0.52 to 6.00) | .36 | 219 | 0.52 (0.28 to 0.97) | .04 |
| 3 | 55 | 2.06 (0.79 to 5.37) | .10 | 239 | 0.52 (0.28 to 0.96) | .04 |
| 4 | 175 | 1.91 (1.10 to 3.32) | .02 | 315 | 0.58 (0.34 to 1.01) | .06 |
| 5 | 93 | 1.92 (0.92 to 4.02) | .08 | 234 | 0.66 (0.37 to 1.19) | .17 |
| 6 | 84 | 1.85 (0.86 to 3.97) | .11 | 318 | 0.54 (0.30 to 0.96) | .03 |
| 7 | 239 | 1.72 (1.05 to 2.82) | .03 | 478 | 0.51 (0.31 to 0.84) | .008 |
Sensitivity analyses: 1 = restricted to DM1 patients with DM1 records in at least two data sources and their matched cohort; 2 = restricted to DM1 patients diagnosed after their clinic’s “up to standard” date and their matched cohort; 3 = Restricted to DM1 patients diagnosed in 1995 or later and their matched cohort; 4 = restricted to patients eligible for linkage to Hospital Episodes Statistics and their matched cohort; 5 = ending follow-up at a maximum of eight years for all patients; 6 = restricted to DM1 patients diagnosed after their registration with their Clinical Practice Research Datalink (CPRD) clinic; 7= excluding cancers identified from the Office of National Statistics only. HR=hazard ratio; CI=confidence interval.
Reported as the total number of cancers in DM1 and DM1-free patients to adhere to CPRD policy, which prohibits the reporting of cells with fewer than five events.
Baseline hazards stratified on matched sets, adjusted for average number of primary care visits per year, and linkage status. Reference group is DM1-free cohort.
Discussion
Using a cohort of 927 DM1-affected patients, spanning the full clinical spectrum of the disease and a matched cohort of 13 085 DM1-free individuals, we showed that excess cancer risk in DM1 was restricted to those with the classic form, driven by previously reported DM1 site-specific cancers including uterine, thyroid, and cutaneous melanoma.
Classic DM1 patients had an approximately twofold increase in overall cancer risk when compared with the DM1-free cohort. This observation is strikingly similar to previously reported findings among the Swedish/Danish (6), Spanish (10), and Utah (9) populations; however, in this study, this finding was observed in classic but not late-onset DM1 patients, suggesting that this more severe disease subset is associated with increased cancer susceptibility. Of note, DM patients included in the Scandinavian and Utah studies were primarily identified from inpatient records, thus likely overrepresenting more severe cases. Interestingly, previous attempts to test this hypothesis using leukocyte repeat size as a marker of disease severity showed no significant associations with cancer (8,10,20,21). It is important to note that the correlations between leukocyte repeat size and age at onset or disease severity in DM1 are only modest (5). In contrast, tumor tissue in DM1 patients had longer repeat expansions compared with surrounding tissue (22–27). The high level of somatic mosaicism related to CTG repeat length in DM1 patients may also explain this lack of association (28). Molecular studies of paired normal and tumor tissues from genetically confirmed DM1 patients are warranted to better elucidate the association between repeat size and cancer risk in DM1.
Excess cancer risk in this study was driven by previously reported affected cancer sites; at least three of the five previous studies reported significant elevated risks of endometrial (6,8–10) and thyroid (6,7,9,10) cancers, as well as cutaneous melanoma, albeit not achieving statistical significance (6–8,10), suggesting that the female genital organs, endocrine system, and possibly skin may be target tissues for carcinogenesis in DM1. Our data also suggested elevated risks of cancers of the ovary, brain, pancreas, and colorectum, all of which were reported as excessive in prior studies (6–10). The molecular mechanism behind DM1 carcinogenesis is still unknown, but data from Fernandez-Torron et al. (10) suggested a possible etiologic role for the microRNA 200c/141 tumor suppressor family. Also, it was previously suggested that altered mismatch repair mechanisms may play a role in the DM1-cancer association based on the close similarities between the cancer profiles of DM1 and Lynch syndrome (6). The observed high relative risk of thyroid cancer in DM1 may argue against this hypothesis because it is not typically observed in Lynch syndrome patients, but several lines of supporting evidence for a mismatch-repair–thyroid cancer association exist, including a few cases in Lynch syndrome patients (29–32).
Our observation of no cancers in congenital/childhood DM1 patients may be a consequence of small sample size, insufficient follow-up (median age at end of follow-up = 25.7 years), or driven by competing mortality in those patients, who may not live long enough to develop cancer. Our results and those of a previous study (33) showed that cancer is an adult phenotype in DM patients (median age at first cancer in this study = 54 years, range = 34–96 years).
Surprisingly, we observed a lower than expected cancer risk in patients with late-onset DM1 compared with their matched DM1-free cohort. These results should be interpreted with caution because an attenuation of effect size was observed in the analysis restricted to patients with a record of DM1 in more than one data source, in which we aimed to reduce potential misclassification of DM1 diagnosis (HR = 0.53, P = .009 in the main analysis, and HR = 0.84, P = .61 in the sensitivity analysis). It is possible that this group of patients may include some patients with the milder DM type 2 whose cancer risk is not yet fully understood. However, it is believed that DM2 accounts for a very small fraction of DM patients in the United Kingdom (∼1–2%) (19,34). Note that the observed reduced risk in this group is driven by deficits of common cancers in an older general population, namely hematological, female breast, lung, and prostate cancers. Lifestyle factors such as smoking and/or body mass index may explain some of these inverse associations; we were not able to adjust for those variables because of missing information.
A major strength of this study is that it used one of the largest primary care medical records databases in the world. This allowed us to evaluate incident cancer in a large DM1 cohort, spanning the full clinical spectrum of the disease compared with a matched DM1-free cohort, thereby minimizing selection bias associated with identifying cases from tertiary care facilities or hospital records, and improving the generalizability of the findings. Studies of CPRD have shown high validity in the reporting of chronic conditions including neoplasms and musculoskeletal disorders (35–37). To optimize the validity of DM1 diagnoses, we conducted multiple sensitivity analyses and our conclusions remained unchanged, demonstrating the robustness of our findings. Nonetheless, our study faced several limitations. It is possible that cancers were more frequently detected in DM1 than in DM1-free patients because the chronic nature of the disease requires more frequent clinic visits. To address this possibility, we adjusted for the average number of primary care visits per year. We also conducted an analysis stratified by calendar time to ascertain potential detection bias in recent years, due to emerging evidence that cancer is part of the DM1 phenotype; we found no heterogeneity. In addition, it is possible that our categorization of disease severity based on age of first record and not actual clinical symptoms resulted in misclassification. To explore this possibility, we conducted a survival analysis of DM1 patients grouped by categories of age at diagnosis compared with DM1-free individuals (Supplementary Figure 1 and Supplementary Table 1). Results showed worse survival in classic DM1 compared with late onset, supporting our classification. Surprisingly, no statistically significant survival difference was observed between patients with congenital/childhood DM1 and those with classic disease. This may suggest that the most severe congenital/childhood DM1 patients may not have been captured in the primary care setting, possibly because of early deaths. Unfortunately, data on repeat expansion size were not available for CPRD patients, thus future studies in genetically well-characterized DM1 patients are warranted. Finally, although large HRs were detected for some cancers, the low frequency of cancer resulted in wide CIs and conclusions about some cancer sites cannot be made.
Our findings in patients with classic DM1 are remarkably consistent with the previous literature, providing further evidence for the increased cancer susceptibility in DM1. The differences in relative cancer risk between classic and late-onset DM1 patients reveal, for the first time, that disease severity modifies DM1-related cancer susceptibility. If confirmed, these novel findings may guide clinical management and scientific planning for investigating the underlying molecular mechanisms in DM-related carcinogenesis.
Funding
The work of Drs Alsaggaf, Pfeiffer, Wang, Greene, and Gadalla was supported by the Intramural Research Program of the Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, US Department of Health and Human Services.
Notes
Affiliations of authors: Clinical Genetics Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Bethesda, MD (RA, YW, MHG, SMG); Department of Epidemiology and Public Health, University of Maryland, Baltimore, MD (RA, DMMSG, MZ, SA); Biostatistics Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD (RMP); Hugo W. Moser Research Institute at Kennedy Krieger Institute, Baltimore, MD (KRW); Departments of Neurology and Neuroscience, Johns Hopkins University School of Medicine, Baltimore, MD (KRW); Marlene and Stuart Greenebaum Comprehensive Cancer Center, University of Maryland, Baltimore, MD (SA).
The authors declare no conflicts of interest.
The authors thank Emily Carver, BS, and David Ruggieri, BS, both from Information Management Services Inc (Calverton, MD, USA) for their important contributions to database management. This study is based on data from the CPRD database October 2016 release, obtained from the UK Medicines and Healthcare Products Regulatory Agency, HES database (© 2016), and ONS database (© 2016) reused with the permission of the Health & Social Care Information Centre. All rights reserved. The interpretation and conclusions contained in this study are those of the authors alone.
The results of this study were selected in part as a platform presentation at the 11th International Myotonic Dystrophy Consortium meeting (San Francisco, September 2017), and as a poster session at the 2018 American Association for Cancer Research Conference (Chicago, April).
Supplementary Material
References
- 1. Bird TD. Myotonic dystrophy type 1. In: Pagon RA, Adam MP, Ardinger HH, Wallace SE, Amemiya A, Bean LJH, eds. GeneReviews®. Seattle, WA: University of Washington; 1993-2018. https://www.ncbi.nlm.nih.gov/books/NBK1165/
- 2. Johnson NE, Heatwole CR.. Myotonic dystrophy: from bench to bedside. Semin Neurol. 2012;32(3):246–254. [DOI] [PubMed] [Google Scholar]
- 3. Brook JD, McCurrach ME, Harley HG, et al. Molecular basis of myotonic dystrophy: expansion of a trinucleotide (CTG) repeat at the 3′ end of a transcript encoding a protein kinase family member. Cell. 1992;69(2):385. [DOI] [PubMed] [Google Scholar]
- 4. Mahadevan M, Tsilfidis C, Sabourin L, et al. Myotonic dystrophy mutation: an unstable CTG repeat in the 3′ untranslated region of the gene. Science. 1992;255(5049):1253–1255. [DOI] [PubMed] [Google Scholar]
- 5. Thornton CA. Myotonic dystrophy. Neurol Clin. 2014;32(3):705–719, viii. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Gadalla SM, Lund M, Pfeiffer RM, et al. Cancer risk among patients with myotonic muscular dystrophy. JAMA. 2011;306(22):2480–2486. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Win AK, Perattur PG, Pulido JS, Pulido CM, Lindor NM.. Increased cancer risks in myotonic dystrophy. Mayo Clin Proc. 2012;87(2):130–135. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Mohamed S, Pruna L, Kaminsky P.. [Increasing risk of tumors in myotonic dystrophy type 1]. Presse Med. 2013;42(9, Pt 1):e281–e284. [DOI] [PubMed] [Google Scholar]
- 9. Abbott D, Johnson NE, Cannon-Albright LA.. A population-based survey of risk for cancer in individuals diagnosed with myotonic dystrophy. Muscle Nerve. 2016;54(4):783–785. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Fernandez-Torron R, Garcia-Puga M, Emparanza JI, et al. Cancer risk in DM1 is sex-related and linked to miRNA-200/141 downregulation. Neurology. 2016;87(12):1250–1257. [DOI] [PubMed] [Google Scholar]
- 11. Wang Y, Pfeiffer RM, Alsaggaf R, et al. Risk of skin cancer among patients with myotonic dystrophy type 1 based on primary care physician data from the U.K. Clinical Practice Research Datalink. Int J Cancer. 2017; 142(6):1174–1181. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Du H, Cline MS, Osborne RJ, et al. Aberrant alternative splicing and extracellular matrix gene expression in mouse models of myotonic dystrophy. Nat Struct Mol Biol. 2010;17(2):187–193. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Osborne RJ, Lin X, Welle S, et al. Transcriptional and post-transcriptional impact of toxic RNA in myotonic dystrophy. Hum Mol Genet. 2009;18(8):1471–1481. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Ranum LP, Day JW.. Myotonic dystrophy: clinical and molecular parallels between myotonic dystrophy type 1 and type 2. Curr Neurol Neurosci Rep. 2002;2(5):465–470. [DOI] [PubMed] [Google Scholar]
- 15. Udd B, Krahe R.. The myotonic dystrophies: molecular, clinical, and therapeutic challenges. Lancet Neurol. 2012;11(10):891–905. [DOI] [PubMed] [Google Scholar]
- 16. De Antonio M, Dogan C, Hamroun D, et al. Unravelling the myotonic dystrophy type 1 clinical spectrum: a systematic registry-based study with implications for disease classification. Rev Neurol (Paris). 2016;172(10):572–580. [DOI] [PubMed] [Google Scholar]
- 17. Herrett E, Gallagher AM, Bhaskaran K, et al. Data resource profile: Clinical Practice Research Datalink (CPRD). Int J Epidemiol. 2015;44(3):827–836. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Allison PD. Survival Analysis Using SAS®: A Practical Guide. 2nd ed. Cary, NC: SAS Institute; 2010.
- 19. Wood L, Cordts I, Atalaia A, et al. The UK Myotonic Dystrophy Patient Registry: facilitating and accelerating clinical research. J Neurol. 2017;264(5):979–988. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Bianchi ML, Leoncini E, Masciullo M, et al. Increased risk of tumor in DM1 is not related to exposure to common lifestyle risk factors. J Neurol. 2016;263(3):492–498. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Das M, Moxley RT, Hilbert JE, et al. Correlates of tumor development in patients with myotonic dystrophy. J Neurol. 2012;259(10):2161–2166. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Mueller CM, Hilbert JE, Martens W, Thornton CA, Moxley RT, Greene MH.. Hypothesis: neoplasms in myotonic dystrophy. Cancer Causes Control. 2009;20(10):2009–2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Bañuls J, Botella R, Palau F, et al. Tissue and tumor mosaicism of the myotonin protein kinase gene trinucleotide repeat in a patient with multiple basal cell carcinomas associated with myotonic dystrophy. J Am Acad Dermatol. 2004;50(2):1–3. [DOI] [PubMed] [Google Scholar]
- 24. Jinnai K, Sugio T, Mitani M, Hashimoto K, Takahashi K.. Elongation of (CTG)n repeats in myotonic dystrophy protein kinase gene in tumors associated with myotonic dystrophy patients. Muscle Nerve. 1999;22(9):1271–1274. [DOI] [PubMed] [Google Scholar]
- 25. Kinoshita M, Igarashi A, Komori T, et al. Differences in CTG triplet repeat expansions in an ovarian cancer and cyst from a patient with myotonic dystrophy. Muscle Nerve. 1997;20(5):622–624. [DOI] [PubMed] [Google Scholar]
- 26. Ogata K, Takahashi A, Oguchi N, Ishitoya J, Fuse S, Shimpo T.. [Somatic mosaicism of p(CTG)n expansion in a case of myotonic dystrophy with parotid tumor]. Rinsho Shinkeigaku. 1998;38(8):736–738. [PubMed] [Google Scholar]
- 27. Osanai R, Kinoshita M, Hirose K, Homma T, Kawabata I.. CTG triplet repeat expansion in a laryngeal carcinoma from a patient with myotonic dystrophy. Muscle Nerve. 2000;23(5):804–806. [DOI] [PubMed] [Google Scholar]
- 28. Ashizawa T, Dubel JR, Harati Y.. Somatic instability of CTG repeat in myotonic dystrophy. Neurology. 1993;43(12):2674–2678. [DOI] [PubMed] [Google Scholar]
- 29. Fazekas-Lavu M, Parker A, Spigelman AD, et al. Thyroid cancer in a patient with Lynch syndrome—case report and literature review. Ther Clin Risk Manag. 2017;13:915–918. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Johnson JM, Chen J, Ali SM, et al. Molecular profiling of synchronous colon cancers and anaplastic thyroid cancer in a patient with Lynch syndrome. J Gastrointest Canc. 2018;49(2):203–206. [DOI] [PubMed] [Google Scholar]
- 31. Gatzidou E, Michailidi C, Tseleni-Balafouta S, Theocharis S.. An epitome of DNA repair related genes and mechanisms in thyroid carcinoma. Cancer Lett. 2010;290(2):139–147. [DOI] [PubMed] [Google Scholar]
- 32. Santos LS, Silva SN, Gil OM, Ferreira TC, Limbert E, Rueff J.. Mismatch repair single nucleotide polymorphisms and thyroid cancer susceptibility. Oncol Lett. 2018;15(5):6715–6726. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Gadalla SM, Pfeiffer RM, Kristinsson SY, Björkholm M, Landgren O, Greene MH.. Brain tumors in patients with myotonic dystrophy: a population-based study. Eur J Neurol. 2016;23(3):542–547. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Norwood FL, Harling C, Chinnery PF, Eagle M, Bushby K, Straub V.. Prevalence of genetic muscle disease in Northern England: in-depth analysis of a muscle clinic population. Brain. 2009;132(11):3175–3186. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Dregan A, Moller H, Murray-Thomas T, Gulliford MC.. Validity of cancer diagnosis in a primary care database compared with linked cancer registrations in England. Population-based cohort study. Cancer Epidemiol. 2012;36(5):425–429. [DOI] [PubMed] [Google Scholar]
- 36. Herrett E, Thomas SL, Schoonen WM, Smeeth L, Hall AJ.. Validation and validity of diagnoses in the General Practice Research Database: a systematic review. Br J Clin Pharmacol. 2010;69(1):4–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Khan NF, Harrison SE, Rose PW.. Validity of diagnostic coding within the General Practice Research Database: a systematic review. Br J Gen Pract. 2010;60(572):e128–e136. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.