Making an impact on mortality in rheumatoid arthritis: Targeting cardiovascular comorbidity
Since the first report in 1953 (1), several studies have confirmed the observation of excess mortality rates in various rheumatoid arthritis (RA) prevalence and incidence cohorts. On review, there appears to be little doubt that people with RA have a lower life expectancy (2–4). These reports, along with data from observational studies demonstrating that most patients eventually experience severe functional decline (5), work disability (6), and joint destruction (7), have led to considerable changes in treatment strategies. Rheumatologists have abandoned the traditional pyramid paradigm (“go low; go slow”) and are treating their patients earlier and more aggressively (8). Introduction of new disease-modifying antirheumatic drugs (DMARDs), more aggressive use of “old” drugs both singly and in combination, as well as reintroduction of corticosteroids in the treatment of early disease have fundamentally changed the approach to treating patients with RA (9). Along with the therapeutic changes, there have been important advances in the standardization and adoption of outcome measures of disease activity, severity, and quality of life (10). Fortunately, these measures are now widely used in almost all trials and observational studies, and we know that therapeutic changes have been successful in improving the quality of life and functioning of RA patients.
These changes in the approach to RA have led to improved morbidity rates, but evidence of improved mortality rates in RA remains relatively scant. For instance, recent cohort studies have documented similar mortality rates in patients with RA and the general population (11, 12), and improved survival associated with response to methotrexate has been reported, but from only 2 treatment centers, albeit with important databases (13, 14). Other studies, however, concluded that survival in RA has not improved in recent decades (15, 16), and one study suggested increased mortality in methotrexate-treated RA patients who have cardiovascular disease (17). Several explanations have been offered for these conflicting signals. First, apparent improvement in survival in certain studies could be due to the more frequent use of population-based cohorts rather than clinic-based or referral cohorts in recent years (18, 19). Second, since the excess mortality in RA does not become apparent until approximately 8–10 years after disease onset, inception cohorts followed too briefly will fail to capture this excess, and any improvement in mortality rates can only be identified after a long lag time (16, 19). Furthermore, there may be hidden problems in retrospective cohort studies that could potentially bias the results (20). Finally, traditional measures of RA disease activity may not necessarily be the best predictors of mortality.
This commentary suggests an additional explanation that rheumatologists should be able to address in the clinic: the undertreatment of—mostly cardiovascular—comorbidity. In very recent work, Navarro-Cano et al (21) confirmed the independent contributions of RA and comorbidity to mortality. Most of this contribution comes from the widespread cardiovascular comorbidity (22). Perhaps excess mortality in RA is primarily linked to etiologies such as the effects of systemic inflammation on vascular integrity (23), and the treatments, although they improve quality of life and functioning, do not improve vascular integrity. Furthermore, evidence suggests that patients with any chronic condition are likely to receive substandard care for other concurrent diseases (24). Thus, undertreatment of comorbidity, especially cardiovascular comorbidity, and underlying risk factors is a matter of concern to all rheumatologists.
Improved morbidity in RA
Evidence is strong that the clinical status of patients with RA is significantly more favorable at this time than in previous decades. The emergence of many early arthritis clinics (and thus, the expanded possibilities for early referral) is a likely explanation for the observation that patients present with less-active disease at the first visit. Callahan et al (22) observed that scores on a modified Health Assessment Questionnaire (M-HAQ) were unchanged over 5 years in a cohort of 100 patients from Tennessee who were monitored from 1985–1990. This finding contrasted with those of all previous studies, which had documented significant declines in function over 5–15 years. Bergstrom et al (25) reported that patients seen in Sweden in 1995 had significantly lower values for C-reactive protein, the Ritchie Articular Index, and Steinbrocker functional class than did patients seen in 1978. This improvement was associated with significantly greater use of DMARDs and glucocorticoids. Sokka et al (26) followed 2 inception cohorts in Finland for 8 years, and found the 8-year radiographic progression in patients starting in 1983 to be half that in patients starting in 1973. This was associated with significantly greater use of single DMARDs and DMARD combinations in the 1983 inception cohort.
A comparison of cross-sectional analyses performed in Tennessee in 1984–1986 compared with 1999–2001 showed all clinical status measures to be significantly better in the latter group (27). Again, this was associated with more extensive use of DMARDs in recent years. In the cohorts resulting from the COBRA (Combinatietherapie Bij Reumatoïde Artritis) trial, both (actively treated) groups had an improvement in their HAQ scores from approximately 1.5 at baseline to 0.7 at 1 year, with a subsequent mean change of 0 after 4 years (28). Finally, da Silva et al (29) documented a decreased need for joint replacement surgery in recent years. These data strongly suggest that through more active and early treatment, morbidity in RA has improved.
Cause of death in RA
A compilation of attributed causes of death in RA patients reported in the world's literature indicated that cardiovascular disease was the most commonly attributed cause of death, being reported in 42% of RA patients (30). This rate was similar to the reported 41% in the 1977 US population. The rates of other attributed causes of death were substantially higher in patients with RA compared with the general population, including infection in 9% of patients (versus 1% of the general population), renal disease in 8% (versus 1%), pulmonary disease in 7% (versus 3%), and gastrointestinal disease in 4% of patients (versus 2% of the general population). There was a reciprocally lower prevalence of cancer, accidents, and “miscellaneous” causes of death in patients with RA. Only 5% of deaths in RA patients were attributed to the RA itself. In some series, no deaths were attributed to the RA. Subsequent observations have yielded similar results, indicating that most deaths in RA are attributed to comorbid conditions. One caveat concerning these data is that death certificates are notoriously unreliable, especially in RA (31, 32).
Markers of premature mortality in RA
Predictive markers such as blood pressure, serum cholesterol, or anatomic stage in Hodgkin's disease are examples of effective measures by which to analyze improvement in mortality rates in other chronic diseases. Traditional measures in RA that are thought to indicate the presence of more severe disease, such as radiographic erosions and rheumatoid factor, are less effective at predicting mortality than are functional status measured by questionnaires or physical functioning measured by grip strength and walking time, as well as the patient's socioeconomic status, comorbid conditions, and age. In most series, the most significant predictor of mortality is either age or functional status as determined by questionnaires such as the HAQ (14).
For example, mortality over 5 years was analyzed in a series of 210 patients with RA in whom careful quantitative measurements were made at baseline in 1985 (4, 22). As might be expected, poorer status at baseline (as indicated by joint counts and radiographic, laboratory, questionnaire, physical function, and demographic measures) predicted decreased survival over the next 5 years. Curiously, the only 3 measures that did not differ statistically significantly at baseline in patients who survived or died over the next 5 years were radiographic scores, presence of rheumatoid factor, and pain scores. In a Cox regression analysis, the 3 most important predictors of mortality were age, functional status, and comorbid conditions. It is of interest that age, comorbidity, and functional status appeared to be the most significant predictors of mortality in patients with other diseases, such as cancer, as well as in geriatric patients.
(Cardiovascular) comorbidity in RA
The analysis of causes of death suggests an increased risk of comorbidity in patients with RA. If cardiovascular disease is the most common cause of death in patients with RA, who die 5–15 years too early, that suggests a significantly increased risk of premature cardiovascular disease in these patients. Indeed, evidence of increased cardiovascular risk in RA is mounting (for review, see ref. 33). A recent study demonstrated the presence of at least 1 comorbid condition in 27% of Dutch patients with RA (34). Not surprisingly, cardiovascular disease constituted 29% of the comorbid conditions. In this inception cohort, RA outcome was not obviously affected by the presence of a comorbid condition. Quality of treatment of the comorbid condition was not the object of study.
In a series of patients with much more longstanding RA (35), 80% had at least 1 significant comorbid condition, including hypertension in 36%, allergies in 22%, stomach ulcer in 21%, chronic bronchitis in 13%, diabetes in 5%, and so forth. The relative odds ratios for the presence of these conditions in RA patients were generally 3–5-fold higher than in the general population without arthritis. These observations may be influenced by Berkson's bias; that is, people who are seen in medical settings are more likely to have diseases identified. However, the prevalence of comorbidities in the Vanderbilt University series was remarkably similar to that in people with symmetric polyarthritis in a population-based US health interview survey of 1978 (36). One need only look in a typical rheumatology clinic to recognize that comorbid conditions are quite common in people with RA.
What are the potential mechanisms for cardiovascular comorbidities in RA? Research on traditional risk factors in RA has shown mixed results. Mobility is decreased in RA patients. There are suggestions, but no unanimity, that cigarette smoking by RA patients is increased compared with that in the general population (37, 38). There were no convincing increases in the prevalence of hypertension or diabetes mellitus that would explain the cardiovascular mortality, and reports on blood lipid levels are also mixed (39–43). Earlier studies revealed little or no evidence of dyslipidemia, but a recent report showed that active RA leads to an unfavorable lipid profile (low cholesterol and very low high-density lipoprotein cholesterol) that improves with treatment (39).
Apart from modification of traditional risk factors for cardiovascular disease by the presence of RA, there are increasing suggestions that RA-specific factors also play an important role (44). It is likely that the first step is injury to the endothelial cells that line the arterial wall. There is considerable interest in the role of inflammation and infection in this process. Recent research has shown that endothelial cell dysfunction not only occurs in the presence of classic risk factors but is also induced by such common pathophysiologic stimuli as bacterial products and inflammatory cytokines, including tumor necrosis factor α and interleukin-1 (44). Ridker et al (45) have shown that C-reactive protein levels in the upper limit of the normal range are the second most-powerful risk factor for acute cardiovascular events. Thus, even in essential atherosclerosis, subclinical inflammation appears to play a critical role (39, 46).
In the group of RA patients who die, the proportion of deaths attributable to cardiovascular disease is similar to that in the general population. Hence, it could be suggested that the interaction between systemic inflammation and traditional risk factors leads to accelerated damage of vessel walls and, thus, to an increased lethality of atherosclerosis in the presence of RA (42, 47). This interaction may start before clinical disease is evident. Nielen et al (48) recently found that RA patients show increases in acute-phase protein levels many years before the onset of clinical disease.
It is important to emphasize that we do not know the extent to which inflammation needs to be controlled in order to eliminate or reduce the inflammation-associated cardiovascular risk in RA. If anything, current morbidity and mortality data suggest that yet more disease control is needed than we are currently reaching in routine care. Thus, efforts at assessing the interplay between systemic inflammatory activity and traditional risk factors should continue, including trials on the effectiveness of pharmacologic interventions for traditional risk factors among RA patients and observational data. In the meantime, existing and updated guidelines for the prevention and treatment of the “metabolic syndrome” of hypertension, obesity, dyslipidemia, and diabetes can also be applied in patients with RA (49–52).
The failure to focus on comorbidity in the treatment of chronic diseases
Several studies have shown that patients with 1 disease may be undertreated for other diseases (53, 54), and RA is no exception (55, 56). This may be partly the result of a confounding effect of age (57, 58); that is, elderly patients are less likely to be treated, and elderly patients are also more likely to have a comorbid condition. The effect is not universal, as a study in breast cancer survivors showed (59). Breast cancer survivors received better-quality noncancer preventive therapy than did non-cancer control patients (taken from a Medicare sample and matched for age, race, and geographic location). However, a pivotal study by Redelmeier et al (24) convincingly showed the negative effect of the presence of one disease on the treatment of other diseases.
Those authors studied all drug prescriptions for the 1.3 million residents of Ontario, Canada, who were over 65 years of age (universal health coverage; drugs free of charge) in 1995. Patients with 3 primary conditions (diabetes mellitus, emphysema, and psychotic syndromes) were classified through specific drug prescriptions and compared with the rest of the population without such prescriptions. For each condition, an unrelated condition was defined that also needed specific drug therapy. The null hypothesis was that having the primary condition would not influence the likelihood of receiving treatment for the unrelated condition. However, compared with the controls, patients with the condition were roughly half as likely to receive treatment for the unrelated condition. Thus, patients with diabetes mellitus were half as likely to receive estrogen replacement therapy than were patients without diabetes. Likewise, patients with emphysema were half as likely to receive lipid-lowering drugs, and patients with psychotic syndromes were half as likely to receive antiarthritic therapy (odds ratios ranged between 0.40 and 0.69).
Several reasons for this effect can be proposed. Patients may be more reluctant to accept additional therapy. Physicians are allotted no more time for patients with complicated conditions and may strive for simplicity of treatment in order to improve compliance and reduce the chances of unwanted side effects and drug interactions. Patients with reduced life expectancy may not benefit enough from preventive therapy to justify extra treatment, or this may be the treating physician's perception. Nevertheless, such undertreatment may have an important impact on prognosis. We propose that this could be an important mechanism to explain the increased mortality in RA.
Treating RA patients early and aggressively has led to measurable improvements in morbidity rates and quality of life, but improvements in mortality rates are much less clear at this time. Interest in cardiovascular comorbidity in RA has increased dramatically in recent years. Better understanding of the interactions between systemic inflammation and blood/vessel wall pathophysiology is essential for guiding the intensity of RA treatment aimed at preventing cardiovascular disease. In the meantime, there is no excuse for ignoring existing guidelines. We propose that a more aggressive approach to state-of-the-art screening and treatment of this extremely prevalent comorbid condition and its well-known risk factors in the rheumatology clinic will lead to immediate improvements in mortality rates, which seem to have eluded us despite important advances in the management of RA.