Introduction
The interplay between cancer and immune-mediated diseases as multiple sclerosis (MS) is intriguing and has been debated for several years. It has been suggested that the abnormal immune response seen in MS could improve surveillance against malignancy [
1]. However, chronic inflammation also represents a recognized risk factors for cancer development [
2].
Studies assessing the prevalence and incidence of cancer in MS reflect this controversy, inconsistently showing similar [
3‐
6], reduced [
7‐
10], or increased risk [
11,
12] as compared to the general population.
Different study designs, methods of case ascertainment and study periods may well explain at least part of such conflicting findings [
13]. Genetics, as well as established MS-associated environmental factors such as smoking, obesity, physical activity, and socioeconomic status may also modulate risk of cancer in patients with MS [
7,
14], acting as relevant confounders when not accounted for. Additionally, the continuous evolution of MS immunomodulatory and immunosuppressive agents possibly influences immune surveillance and cancer development in MS. Finally, only few studies have included MS patients diagnosed with MS during the last decade and no clear temporal trends in cancer diagnoses among MS patients has emerged [
13‐
17].
We, therefore, aimed at investigating in primary care settings the occurrence of cancer in MS patients as compared to matched controls from the general population, and how this has evolved over time in the last 25 years, using the United Kingdom (UK) Clinical Practice Research Datalink (CPRD) [
18]. As an exploratory aim, we also estimated cancer occurrence in MS patients and paired controls before a diagnosis of MS is made.
Discussion
We observed that individuals who received a diagnosis of MS between 1990 and 2016 in the UK had an overall comparable occurrence of cancer diagnoses as compared to age, sex, and general practitioner matched subjects. A trend for a lower incidence of cancer among MS patients was actually present, but did not reach statistical significance. These results are in line with those of several studies from Northern Europe and Iran, reporting no differences in cancer risk between MS patients and the general population [
3,
5,
15,
22]. Several other studies from the US, Canada, and Europe instead, found this risk to be overall decreased [
7‐
9,
13,
17]. These findings appear even stronger in light of the expected risk of surveillance bias among individuals suffering from a chronic condition such as MS [
4]. Interestingly, a recent study from Norway found an increased risk of cancer among MS patients, with a particular involvement of the respiratory, urinary, and central nervous systems [
11]. In our MS cohort, however, no signals for any specific type of cancer were detected. Taken together, we can conclude that MS patients in the UK appear to be at overall similar (if not lower) risk of malignancies as compared to the general population, and this is in line with the majority of reports on this topic [
3‐
6].
A recent systematic review of cancer incidence and prevalence in the MS population showed a significant variability and inconsistencies among results [
23]. Such conflicting findings are likely related to a variety of confounding factors including different study designs, methods of data collection, time periods of the studies, differences in cancer screening programs across countries, and last but not least population-specific genetic and environmental exposures.
We noted that earlier and recent studies have generally assessed cancer occurrence irrespective of the epoch of MS diagnosis. Only one study from Sweden investigated the relationship between cancer risk and the calendar year of study entry, showing similar cancer rates in MS patients diagnosed between 1969–1980 vs 1980–2005 [
7]. However, categorizing time according to pre- vs post-1980 appears rather arbitrary and the most recent years (when the majority of new immunosuppressive therapies have become available) were also not included. We, therefore, aimed at investigating a potential change in cancer occurrence across time. Interestingly, we found that, while overall cancer risk was stable over time in the control group, it consistently increased among MS patients by approximately 2% per calendar year at index date (date of a first MS code in CPRD).
It is not easy to explain these findings. It is intriguing to hypothesize that the abnormal immune response seen in MS patients may exert a protective effect against cancer development through increased immune surveillance. This may explain the historically comparable or even reduced rates of cancer among MS patients, despite the likely surveillance bias and the presence of risk factors common to MS and cancer such as smoking and obesity [
24,
25]. Our results suggest, however, an apparent change in cancer diagnoses among MS patients in the UK with a variety of possible explanations. Several changes have definitely occurred between 1990 and 2016 in the field of MS, among those the introduction of several new potent therapeutic agents (many with a strong immunosuppressive effect), changes in clinical care, standardized and regular cancer screening programs and changes in lifestyle behaviours [
26].
If a protective effect of MS was present against cancer, one might expect it to precede MS symptoms due to ongoing subclinical pathological immune processes, or intrinsic genetically and/or environmentally determined individual factors influencing both MS and cancer risk. To investigate this hypothesis, we also assessed cancer risk in MS individuals at 0–2, 2–5, and 5–10 years before index date, finding no differences as compared to matched controls. Similarly, a previous study by Fois et al. based on UK hospital admissions between 1963 and 1999, found no increased risk of cancer, irrespective of whether MS diagnosis preceded or followed cancer diagnosis [
22]. Thormann et al. also found a similar cancer risk before a first record of MS in 8,947 Danish MS patients diagnosed between 1980 and 2005, as compared to matched individuals [
17]. Despite the limited number of studies investigating cancer risk before MS diagnosis, taken together these findings suggest that any possible protective effect of MS against cancer does not anyhow appear before a definite diagnosis of MS is made [
17,
22]. A caveat should be, however, mentioned, that the power of these studies might be insufficient to detect any differences due to the lower number of cancer events in younger individuals.
Our study has several limitations. Electronic medical records represent invaluable tools able to provide sample sizes that are large enough to investigate associations of small effect and subtle changes in disease rates over time. This, however, does not come without problems. MS as well as cancer diagnoses were identified when the first respective code was reported in CPRD, but this does not necessarily reflect the year of diagnosis. This is even more complicated by the fact that MS diagnostic criteria have been revised several times during the study period. Validation studies have, however, been performed supporting the reliability and quality of CPRD data and CPRD-coded diagnoses, including MS [
19,
20]. Moreover, we could replicate some known associations, i.e. increasing risk of cancer with age, and increasing risk of specific cancer subtypes such as lung cancer with male gender, making our results more reliable. Second, we do not provide any data concerning possible confounding factors, such as lifestyle behaviours, sun exposure, smoking, social status, body weight and mostly disease modifying therapies, whose potential role cannot be disentangled in this context. Cases and controls were matched by GP practice, which is an indirect indicator of geographical area. Despite the evidence for a good correlation between area of residence and socioeconomic status [
27], individual measures of socioeconomic status were not available and imbalances in the matching of cases and controls in this regard are possible. We also did not attempt to investigate the effect of specific treatments on cancer risk, as information regarding drugs and infusions prescribed by the treating neurologist may be absent or incomplete in a primary care database such as the CPRD.
In conclusion, we showed no significant differences in occurrence of cancer in the UK between MS patients and the general population. However, we highlight a mild progressive increase in cancer diagnoses among patients with a first record MS between 1990 and 2016, a finding that requires further investigations. It would be particularly interesting to see whether similar changes have indeed occurred in other countries than the UK. While several explanations appear to be possible, including increasing surveillance and more careful cancer screening programs in MS patients, we believe the rapid and continuous evolution of MS care, treatments and related potential secondary effects, require maximal attention in routine neurological care.
Compliance with ethical standards