Time Trends in Epidemiology and Characteristics of Psoriatic Arthritis Over 3 Decades: A Population-based Study

DISCUSSION

We are the first to report on the time trends in the incidence and prevalence of PsA using the CASPAR criteria over a 30- year period in a geographically defined population. Our findings indicate that there is a progressive increase in the incidence of PsA between 1970 and 2000. We have estimated that by the year 2000, the age- and sex-adjusted incidence rate for PsA was 10 per 100,000, with a prevalence of 158 per 100,000. Using these findings, we have extrapolated that between 162,000 to 589,000 adults aged ≥ 18 years were affected with PsA in the US in 2000, and 8,000 to 27,000 new cases occur each year.

There are only a limited number of incidence studies of PsA^1,3,6,7,12. Ours is the first study that examines PsA incidence trends using the same classification criteria consistently over an extended period. Although our PsA incidence estimates are broadly consistent with previous estimates, the most significant finding is the increasing incidence over time. This is in contrast to RA, a Th1-mediated disease, which has shown a decline in incidence over the same period¹⁹. Previous studies have suggested a birth-cohort effect to explain the decline in RA, which had a higher incidence in the beginning of the 20th century, suggesting infectious or environmental triggers²⁰. Reasons for the change in PsA incidence are also unclear, but may be due to greater physician awareness, the changing nature of the disease, or unknown environmental or genetic risk factors^21–24. Of note, the incidence of psoriasis also has increased in this population during the same period^25,26, and thus the increase in incidence of PsA may parallel the increasing psoriasis trend.

Intriguingly, our findings also demonstrate a difference in the age-specific incidence of PsA between men and women. Contrary to prevailing belief that PsA occurs equally in both sexes, our findings indicate that the incidence of PsA in women is less than the incidence in men until the sixth decade of life^8,12. A similar pattern was observed in the incidence of psoriasis in the Olmsted County population and in a recent database study from the United Kingdom^25,27. These findings suggest a possible hormonal influence in the onset of PsA. Pregnancy has been associated with a reduced likelihood of PsA²³. Further, PsA improves during pregnancy, but disease flares are common during the postpartum period when estrogen levels are in flux^28–31. Also, patients treated with estrogen-modifying drugs were less likely to develop PsA²¹. Although evidence is limited in PsA, research in RA and systemic lupus erythematosus has demonstrated that changes in cortisol, estrogen, progesterone, and glucocorticoids levels and the androgen/estrogen balance result in a downregulation of the Th1 response and a shift toward a Th2 cytokine profile³². In summary, our sexspecific incidence patterns provide indirect evidence of the potential role of sex hormones in the etiology of PsA. Further research is needed to elucidate the complex role of sex hormones in the pathogenesis of PsA in men and women.

Our findings mostly agree with Shbeeb, et al, who examined the incidence and prevalence of PsA within the Olmsted County population between 1982 and 1991 using the resources from the REP. They found an overall incidence of 6.59 per 100,000 and a point prevalence of 1 per 1000. This is very similar to our 1980–89 estiamte of 7.0 per 100,000. In contrast to our study, the PsA case definition used by Shbeeb, et al excluded subjects without psoriasis. In our study, we identified 30 subjects who had no evidence of psoriasis and yet fulfilled CASPAR criteria based on family history of psoriasis. Further, the clinical characteristics of the PsA cohort reported by Shbeeb, et al are similar to ours, with an average age at diagnosis of 40.7 years, and the majority of subjects presenting with oligoarthritis as well. One notable difference is sex differences in incidence. In that earlier cohort, the incidence of PsA was approximately equal in both sexes, while we found a higher incidence in men than in women. However, we had the advantage of examining trends over a 30-year period, which may also account for the sex difference. Another significant study examining the prevalence of PsA in the US population is the 2001 telephone survey conducted by the National Psoriasis Foundation³³. In that cross-sectional, self-reported survey, the overall point prevalence of PsA was 0.25%, reaching 11% among patients with psoriasis. Despite the differences in ascertainment methodology, this estimate based on self-reports is surprisingly similar to ours²⁶.

Our study also provides further evidence on the clinical and radiographic features in early PsA. First, most subjects in our incidence cohort satisfied CASPAR criteria based on the presence of psoriasis and a negative RF. This is consistent with patterns seen at other centers¹⁵. Second, our study demonstrates an oligoarticular pattern with predominance of enthesopathy, similar to other early PsA case series^34,35. In contrast, studies of prevalent PsA subjects have documented the predominance of polyarticular involvement^36–38. As suggested by Gladman, et al, PsA begins as an oligoarticular disease that may evolve into a polyarticular phenotype³⁹. Third, a common radiographic feature in our incidence cohort was erosions, especially at the distal interphalangeal joints. Uncommon findings were osteolysis, fluffy periostitis, and bony proliferation or ankylosis. Potential reasons for less severe radiographic features in this inception cohort include underrecording of certain radiographic features by the general radiologists or ascertainment of subjects during early stages of PsA. Most previous studies included prevalent PsA subjects who may have prolonged exposure to the proinflammatory cytokines that are responsible for promoting the cascade that leads to bone destruction and new bone formation^24,40. As demonstrated^41–43, development of radiological damage is related to duration of clinically evident joint inflammation, and the number of actively inflamed joints is associated with the progression of radiological joint destruction. Alternatively, erosive disease or severe radiographic changes of bony deformation are more frequent in PsA subjects who have symmetrical, polyarticular disease, a group not common in our study^8,44,45.

In contrast to previous studies^46,47, we did not observe an increase in overall mortality in this cohort over an average followup of 13.6 years. This was also reported by an earlier study in this population. During the 30 years of the study, the underlying Olmsted County population was stable, without any major changes in demographics or major population shifts. As recently addressed by Ali and colleagues⁴⁸, one of the reasons for the discrepant findings may be an early, less severe stage of the disease in our incidence cohort, with the potential for exposure to sustained treatment that dampens the inflammatory state in later years. In other words, subjects with severe PsA are underrepresented in our inception cohort, whereas they are overrepresented especially in the earlier years of the Canadian clinic-based cohort, where the average PsA disease duration was 4.5 years⁴⁸. Alternatively, our PsA inception cohort is not large enough to detect a small excess in mortality. Indeed, our confidence intervals indicate that, if there is a small increase in risk of death, we cannot exclude a risk less than 37%.

Our study has several potential limitations. First, because of the variable disease course, there is the possibility that we missed PsA cases that presented with minimal or no psoriatic skin lesions, as inflammatory arthritis may precede the diagnosis of PsA in roughly 15%–19% of subjects^37,49. The sensitivity of the CASPAR criteria is only 91% and therefore, some true-PSA cases may not have fulfilled the CAS-PAR criteria. However, over the course of 30 years, we expected that most PsA cases and psoriasis lesions would have been ascertained due to the comprehensive medicalrecords linkage system provided by the REP¹⁶. Also, PsA subjects who never seek medical care would be missed in our study, leading to an underestimation of the true incidence of PsA. We assume that most patients with PsA will be diagnosed, albeit with a delay, during the 30-year study period. In addition, the CASPAR criteria were validated in hospital patients with inflammatory arthritis and not tested yet in a community-based setting. Community-based physicians may not be as vigilant in documenting clinical findings specific to PsA such as dactylitis. Nevertheless, referral to a rheumatologist is relatively easy in the Mayo health system and clinical findings missed on initial examinations would be documented by the rheumatologists and/or dermatologists.

Second, there is the concern of reviewer/detection bias due to the retrospective design of the study. However, the majority of data abstraction was performed with a standardized, pretested data abstraction form that included explicit variable definitions provided by the CASPAR criteria¹⁴. Questions regarding individual PsA diagnosis were discussed with the other coinvestigators and resolved by consensus.

Third, the availability of information in the medical records may be subject to clinician variability; information on clinical variables, such as nail pitting, dactylitis, or the number of affected joints, may have been incompletely documented. Hence, it is plausible that the 128 subjects excluded may have had PsA (especially 40 with psoriasis), but documentation was inadequate in the medical record to have satisfied the CASPAR classification criteria. Similarly, radiographic results in the earlier years of the study were available only as paper copies with reports of “erosions typical of psoriatic arthritis” without further elaboration. Hard copies of radiographs were not reviewed for detailed confirmation of the presence or absence of individual radiographic features. Nevertheless, radiographic evidence was not necessary if a patient already accumulated 3 points to achieve the CASPAR criteria. Further, patients diagnosed with psoriasis prior to arthritis and treated with longterm methotrexate for the skin disease may have had suppression of an inflammatory arthritis, thus underrepresenting PsA in this population. In addition, although we found no difference in survival among subjects with PsA than in the general population, it is possible that not enough subjects were studied for an adequate period.

Finally, the population of Olmsted County is primarily White (90%) and thus limits the generalization of our incidence estimates to racially diverse populations.

Nevertheless, strengths of our study include the unique medical-records linkage system provided by the REP that allows near-complete ascertainment of all clinically recognized cases of PsA in a well defined population. In addition, ours is the first study that examines not only the time trends in the epidemiology of PsA, but also relies on the newly validated CASPAR criteria, which have been shown to have a sensitivity of 91.4% and sensitivity of 98.7%¹⁴. The CASPAR criteria thus can be used in retrospective studies, provided that documentation of clinical and radiographic characteristics are available.

Our findings indicate that the incidence of PsA has been rising over 30 years in men and women. Incidence patterns in men and women follow different patterns, suggesting the potential role of sex hormones in the development of PsA. Further research is needed to elucidate the reasons for the increase in PsA incidence and the sex-specific patterns, including the potential role of environmental and hormonal risk factors.