The impact of clinical symptoms and endoscopic and histologic disease activity on health-related quality of life in patients with ulcerative colitis following treatment with multimatrix mesalazine

Ulcerative colitis (UC) is a chronic, recurrent inflammatory bowel disease (IBD) characterized by inflammation of the colon and rectum. Health-related quality of life (HRQoL) is a multidimensional concept that captures impacts of a health condition, and its treatment, on an individual’s daily physical, emotional, mental, and social functioning, as well as the impact of the individual’s perceived health on their ability to live a fulfilling life [1, 2]. In recent years, HRQoL has been recognized as an important outcome when evaluating health, and improvements in health due to treatment and quality of care, in patients with UC [3]. When in remission, patients with UC are typically asymptomatic, with HRQoL similar to the general population [4‐7]. However, patients with active UC often experience fatigue, increased need to defecate, diarrhea, bloody stools, and abdominal pain. HRQoL is typically sub-normal in patients experiencing acute flares [5, 8] while effective treatment of patients with active UC has been shown to improve HRQoL [9‐13].

The effectiveness of treatment for patients with active UC in clinical trials is typically evaluated by the likelihood of inducing clinical remission and mucosal healing. Clinical remission is indicated as a substantial reduction of clinical symptoms, while mucosal healing reflects a sizable decrease in inflammation and ulceration of colonic and rectal tissue, as evidenced from endoscopic and histologic assessments [14]. In clinical practice, evidence for patients’ treatment response has typically focused on clinical symptoms. Following evidence showing that inflammation in the mucosa is associated with an increased likelihood of relapse [15‐18] as well as increased risk of developing colorectal cancer [19‐21] and colectomy [22, 23], there has recently been more emphasis on mucosal healing as a goal of UC therapy, accompanied by calls for including endoscopic and histologic endpoints in clinical trials [24‐29].

While previous studies have established that increased HRQoL for patients with UC is predicted by reduced clinical symptoms [30‐32] as well as improved mucosal health [33‐37], none of these studies assessed whether improvements in clinical symptoms and mucosal healing are uniquely associated with improvements in HRQoL. Findings regarding the independence of clinical symptoms and mucosal health are mixed: some studies report very strong correlations between the two [38‐40], but others report marked dissociations, with numerous patients in clinical remission who do not exhibit mucosal healing, and vice versa [41‐43]. Given these inconsistencies, it is unclear whether changes in HRQoL are uniquely associated with changes in both clinical symptoms and mucosal health. The current study examines whether changes in clinical, endoscopic, and histologic disease activity are uniquely associated with meaningful changes in HRQoL for patients with UC following disease treatment.

Descriptive statistics for patients’ baseline characteristics for the full induction phase efficacy population (n = 717) are presented in Table 1. All mean scores on UC-DAI components ranged between 1 and 2 points, indicating mild-to-moderate clinical and endoscopic symptoms. Mean SF-12v2 scores were below the general population average (i.e., < 50) on all domains, with deficits particularly large for domains capturing social functioning, role limitations, and bodily pain.

Correlations between UC-DAI component scores and SIBDQ domains (Table 2) were weak to moderate, ranging from − 0.18 (between mucosal appearance and systemic symptoms) to − 0.47 (between stool frequency and bowel symptoms). Correlations across all SIBDQ domains were generally largest for the stool frequency component and smallest for mucosal appearance. The SIBDQ bowel symptoms and emotional function domains showed the largest associations with UC-DAI scores, both with moderate correlations with component scores and strong correlations with UC-DAI total score, while the systemic symptoms domain was the least associated with UC-DAI scores. Correlations between UC-DAI component scores and SF-12v2 domains (Table 2) were also weak to moderate, ranging from − 0.16 (between mucosal appearance and vitality) to − 0.40 (between stool frequency and role physical). Following the same pattern observed for SIBDQ domains, the magnitudes of correlations across all SF-12v2 domains were generally largest for the stool frequency component and smallest for mucosal appearance. Moderate correlations with UC-DAI components were observed for bodily pain, social functioning, role physical, role emotional, and general health domains, while weak correlations were observed for mental health, vitality, and physical functioning domains. Changes in TOGS histology scores were weakly correlated with changes in all SIBDQ domains and all SF-12v2 domains.

All SIBDQ and SF-12v2 domains were responsive to markers of improvement in disease activity, as indicated by statistically significant mean differences (all Hochberg-adjusted p < 0.001) between patients who achieved clinical or endoscopic remission or ≥ 1-point improvement in stool frequency or rectal bleeding from baseline to final study visit and those who did not (Table 3). Overall, SIBDQ domains were more responsive than SF-12v2 domains to all subgroup differences, with the distinction most prominent for achievement of mucosal healing, which had a large overall impact on SIBDQ domains (average d = 0.95) and a medium-sized impact across SF-12v2 scores (average d = 0.72). The SIBDQ bowel symptoms domain was consistently most responsive to the absence or presence in improvements in disease activity, while the systemic symptoms domain was consistently least responsive. Among SF-12v2 domains, general health, bodily pain, and social functioning were consistently most responsive to marker-based improvements in disease activity, while vitality, physical functioning, and mental health were consistently least responsive.

Scores on SIBDQ and SF-12v2 domains at patients’ final visit varied as a function of their histologic disease activity at that visit, as indicated by statistically significant mean differences (all Hochberg-adjusted p < 0.05) between subgroups of patients classified by the cutoff values of 3.1, 4.1, and 5.1 (Table 4). The largest differences in SIBDQ and SF-12v2 mean scores were observed between subgroups based on the highest cutoff value (5.1 points); differences in SIBDQ and SF-12v2 domains were generally comparable between subgroups defined by cutoff values of 3.1 and 4.1. Consistent with findings across clinical and endoscopic subgroups, SIBDQ domains were generally more responsive to patients’ histologic disease status than SF-12v2 domains. Again, the SIBDQ systemic symptoms domain was consistently least responsive to disease activity; responsiveness was similar among the other 3 SIBDQ domains. Among SF-12v2 domains, social functioning was consistently most responsive to histologic disease activity, with general health consistently the least responsive.

Comparisons of mean change in scores from baseline to final visit among MH+/CR+, MH+/CR−, and MH−/CR− subgroups are presented in Fig. 1 for SIBDQ domains and Fig. 2 for SF-12v2 domains. For all domains of both instruments, patients who achieved both mucosal healing and clinical remission or mucosal healing alone showed statistically larger improvements than patients who achieved neither mucosal healing nor clinical remission (all Hochberg-adjusted p < 0.001). Further, for all domains of the SIBDQ and for 6 of the 8 SF-12v2 domains (all but physical functioning and mental health), patients who achieved both mucosal healing and clinical remission scored better than patients who achieved mucosal healing but not clinical remission (all Hochberg-adjusted p < 0.05).

Findings from multivariable linear regression models are presented in Table 5. Patients’ baseline characteristics (age, gender, and BMI) were not significant predictors of changes for the majority of SIBDQ and SF-12v2 domains. Baseline values for each domain were strong and statistically significant (all p < 0.001) predictors of decreased scores, such that patients with worse values at baseline showed greater improvements following treatment. Each of the 4 UC-DAI components accounted for a statistically significant amount of unique variance for the majority of HRQoL domains. For the SIBDQ, all 4 components were significantly associated with bowel symptoms and emotional and social function domain scores, and all but mucosal appearance and physician’s global assessment were significantly associated with change in scores on the systemic symptoms domain. Across SIBDQ domains, regression weights for stool frequency (βs ranging from − 0.113 to − 0.187) and physician’s global assessment (βs ranging from − 0.087 to − 0.181) were generally numerically larger than for mucosal appearance and rectal bleeding. Across UC-DAI components, regression weights were generally numerically largest for changes in bowel symptoms (βs ranging from − 0.109 to − 0.187) and generally numerically smallest for changes in systemic symptoms (βs ranging from − 0.038 to − 0.113). For the SF-12v2, changes in stool frequency were associated with statistically significant changes in all domains, while changes in each of the other 3 components were associated with significant changes for at least half of the domains (role physical, bodily pain, general health, and vitality for rectal bleeding; role physical, bodily pain, general health, social functioning, and role emotional for mucosal appearance; and role physical, bodily pain, social functioning, role emotional, and mental health for physician’s global assessment). Across SF-12v2 domains, regression weights for stool frequency (βs ranging from − 0.083 to − 0.177) were generally numerically larger than for the other UC-DAI components (βs ranging from − 0.044 to − 0.164). Across UC-DAI components, regression weights were generally numerically smallest for changes in physical functioning, vitality, and mental health domains (βs ranging from − 0.047 to − 0.120). Change in patients’ TOGS histology score did not account for significant variability in any SIBDQ or SF-12v2 domains.

Researchers have recently suggested that assessment of disease activity of patients with UC by clinicians should incorporate evaluations of 4 factors: clinical symptoms, endoscopy, histology, and HRQoL [54].

Historically, in the 1980s and 1990s, treatment benefit was defined as improvement or remission in clinical symptoms [55, 56]. During the first decade of the 2000s, expert consensus was reached that classification of treatment response or remission, both in clinical trials and in practice, needed to also include direct evidence of mucosal healing as measured by endoscopy [14, 56‐58]. The importance of including mucosal healing when assessing patients’ disease activity has been indirectly supported by evidence that induction of clinical and endoscopic remission is strongly associated with improvements in HRQoL [31, 33, 36, 59]. Within the past decade, many researchers have claimed that establishing “complete” or “deep” remission in patients with UC requires histologic remission in addition to clinical and endoscopic remission, and that this should be a target therapeutic goal [26‐29]. However, to this point, the contribution of histologic healing and HRQoL has not been thoroughly examined in patients with UC.

Findings from the current analyses support previous evidence that both improved clinical symptoms and mucosal healing were associated with improvements in HRQoL for patients with active mild-to-moderate UC. Changes in scores on each component of the UC-DAI from baseline to patients’ final visit were moderately correlated with changes in scores on most SIBDQ and SF-12v2 domains, particularly bowel symptoms, emotional function, and social function domains of the former, and role–physical, bodily pain, general health, social functioning, and role–emotional domains of the latter. Patients who displayed meaningful improvements in key clinical and endoscopic activity markers, such as achievement of clinical or endoscopic remission, had substantially larger improvements in SIBDQ and SF-12v2 domain scores than those who did not, as did patients who achieved both endoscopic and clinical remission compared to patients who achieved endoscopic remission only. In addition, patients with histology indicating active disease at the end of treatment scored significantly worse on all SIBDQ and SF-12v2 domains than patients with no inflammation. Finally, decreases in each of the UC-DAI component scores, even when controlling for variance shared with the other three UC-DAI components, were statistically significantly correlated with improvement in scores for the majority of SIBDQ and SF-12v2 domains, though change in histology (TOGS) was not correlated with changes in HRQoL outcomes.

Patients who achieved clinical remission, who demonstrated full mucosal healing, or who showed meaningful reductions in stool frequency or rectal bleeding reported significantly larger improvements in all HRQoL domains than did their counterparts. However, among each of these factors, achievement of mucosal healing produced the largest subgroup differences for HRQoL improvements. The driving factor behind this larger difference was not that achieving mucosal healing led to greater improvement than did meeting other markers: across all markers, the mean change in domain scores were relatively comparable for those showing improved disease status. Rather, the distinction of mucosal healing from the other factors is most apparent when examining those who failed to demonstrate improvement. Patients who did not achieve clinical remission still demonstrated some improvement in HRQoL, with increases on SIBDQ domains ranging from 1.4 points (systemic symptoms) to 3.2 points (bowel symptoms) and on SF-12v2 domains from 2.6 points (physical functioning) to 5.8 points (bodily pain). In contrast, patients who did not exhibit endoscopic remission at their final visit showed on average only trivial improvements in HRQoL, with increases < 1 point on all SIBDQ domains and < 2 points on all SF-12v2 domains. Taken together, these findings indicate that mucosal healing, unlike symptom reduction, may be a necessary condition for improvement of HRQoL for patients with UC.

Results from regression models provide further support for the independent contribution of clinical symptoms and mucosal health to HRQoL. Each component of the UC-DAI was uniquely associated with significant variability in the majority of SIBDQ and SF-12v2 domains, indicating that clinical and endoscopic activity of patients with UC each have at least some distinct, and thus additive, associations with their HRQoL.

Results from both correlation and regression models found a lack of linear relationships between changes from baseline to final visit in histology score (TOGS) and changes in SIBDQ or SF-12v2 domains. However, post-treatment comparisons of HRQoL across groups classified by predefined markers of histologic disease activity indicated that histologic status was statistically associated with patients’ concurrent HRQoL. Because these two analyses differed in both how histology was scaled (as a continuous or categorical variable) and whether using current histology status or change in histology status, a post hoc analysis calculated Spearman correlations between TOGS at the final visit with concurrent HRQoL scores. Magnitudes of all correlations were very small, suggesting that differences in HRQoL as a function of histologic disease activity may vary as a function of kind, as defined by clinically meaningful categories, rather than degree.

Across all analyses, some consistent patterns emerged with respect to the relative strength of associations among specific dimensions of disease activity with HRQoL. One such pattern of findings pointed to changes in stool frequency being more related to changes in HRQoL than were changes in rectal bleeding. Median correlations across domains were slightly larger for changes in stool frequency than rectal bleeding for both the SIBDQ and SF-12v2. Differences in HRQoL for patients achieving improvement as compared to those who did not were also larger for stool frequency than rectal bleeding, with average effect sizes across SIBDQ and SF-12v2 domains being larger for the comparison of stool frequency subgroups than for the comparison of rectal bleeding subgroups. Finally, in all but one of the multivariable regression models (the exception being the model for the general health domain of the SF-12v2), the standardized regression weight for change in stool frequency was of greater magnitude than that for rectal bleeding.

The analyses reported here included data from assessments for only the induction phase of the trial, and not for the maintenance phase. The rationale for this decision was related to the established finding that detecting associations among changes in outcomes, which is the focus here, is impacted by the amount of variability in these changes, such that a restricted range of values among variables attenuates their intercorrelations [60]. As would be expected, during the induction phase, when patients with active UC (and thus poor clinical and HRQoL outcomes) received treatment, the majority of patients showed improvement in these outcomes, with large variation in the magnitudes of change across patients [61]. However, also as would be expected, in the maintenance phase, which enrolled only patients in partial or full remission who then continued to receive treatment, changes in clinical and HRQoL outcomes were observed for a minority of patients, with little variation in the magnitudes of change across patients [44, 61]. This restricted range of values in the maintenance phase would likely underestimate the associations among changes in clinical and HRQoL outcomes. Thus, it was determined that the objectives of the current analyses would be best served by including data from only the induction phase of the MOMENTUM trial.

One limitation of the current clinical trial is the use of shortened instruments for assessing patients’ disease activity and HRQoL, which restricts the precision of measurement for each of these outcomes. While the UC-DAI and similar indices capturing both clinical and endoscopic activity in patients with UC (e.g., Mayo score [62]) are frequently included as efficacy endpoints in clinical trials, they are by design simplified measures of these outcomes. In particular, the UC-DAI is limited both in the scope of health outcomes measured—only two clinical symptoms, and no assessment of histologic or biochemical disease activity—and in the variability of possible outcomes due to all items being scored on a highly compressed 4-point scale. An instrument that captures additional clinical symptoms and uses an expanded scale to capture finer distinctions in disease activity could improve the breadth and precision of estimates of patients’ disease activity, leading to more accurate assessments about their association with other variables, such as HRQoL. Similarly, the SIBDQ and SF-12v2 are, by design, simplified measures: each was developed as a subset of items from a more comprehensive scale, specifically the IBDQ and the SF-36v2^® Health Survey (SF-36v2). While previous studies of patients with UC have provided evidence supporting the sensitivity to change and construct validity in UC samples of the SIBDQ [13, 49, 63, 64] and SF-12v2 [10, 32], using their parent instruments would provide more precise estimates of patients’ HRQoL. Further, while responder definitions, or thresholds indicating clinically meaningful change, have been established for scores on both the IBDQ and the SF-36v2 [65, 66], they have not been established for domain scores of the SIBDQ or SF-12v2. Thus, we cannot make inferences from these results as to whether the magnitudes of changes observed are clinically meaningful or relevant.

Another limitation of this study is the inability to examine how disease activity and HRQoL covary over time during treatment. It is possible that the relative magnitudes of associations between clinical symptoms or mucosal health and HRQoL (or, perhaps, with specific aspects of HRQoL) change over time. For example, it may be that clinical symptom reduction has a strong impact on HRQoL early in treatment, while the impact of mucosal healing of HRQoL appears later in the course of treatment. However, due to the lack of interim assessments of HRQoL during the induction phase of the MOMENTUM trial, changes in associations across different durations could not be assessed. While it may appear that including HRQoL data from patients during the maintenance phase of this trial would allow for comparing the magnitudes of associations at multiple assessments, this would in fact not be the case, for two reasons. First, patients who entered the maintenance phase were only a subset of patients who completed the induction phase—specifically, those who achieved partial or complete remission. Thus, it would not be possible to compare associations at week 8 of the induction phase with those at the end of the maintenance phase, as the patient samples for each of the two phases were not the same. Second, the maintenance phase, while much longer than the induction phase (12 months vs. 8 weeks), also did not include interim assessments of HRQoL. Thus, trials incorporating multiple assessments of these variables would be needed to provide important information about the development of these relations over the time course of the disease and its treatment.

The inclusion of only patients with mild-to-moderate UC in this study sample limits the ability to generalize the current findings to the full patient population, which also includes patients with more severe disease. Future research that examines the association between clinical, endoscopic, and histologic activity with HRQoL for patients with severe UC would be needed to understand whether the current findings are applicable to patients with UC across the entire range of disease severity.

Despite these limitations, results from this study, which is the first to simultaneously examine the individual impact of clinical, endoscopic, and histologic activity on HRQoL for patients with UC, provide clear evidence that clinical symptoms and mucosal health have separable, distinct, additive impacts on HRQoL of patients with UC. Treatments that target both clinical and mucosal health will likely result in greater improvements in patients’ HRQoL than those that are directed at only one or the other. Thus, these findings are supportive of the recently recognized importance of using evidence for “complete” remission when evaluating the effectiveness of treatment in clinical trials and hopefully to some extent in clinical practice. Achieving “complete” remission of UC, which includes histologic remission in addition to both clinical remission and mucosal healing, contributes to the ultimate therapeutic goal of treatment for patients with UC, which is to improve HRQoL [27].