The History of the Combination Therapy Versus Monotherapy Debate (How Did We Get Here Anyway?)

With the advent of anti-TNFα therapy in the 1990s with infliximab (IFX), practitioners had at their disposal a rapidly acting and highly efficacious therapy for both luminal and perianal Crohn disease (CD). However, because of the significant cost of these therapies and a less clear understanding of the benefits of scheduled maintenance therapy, episodic use of IFX was common both in clinical practice and in clinical trials. It soon became apparent that the episodic use of IFX was associated with increased immunogenicity as measured by higher levels of antibodies to IFX (ATI) ( Table 1 ). Rates of ATI were as high as 75% in one study. In a study by Maser and colleagues, episodic use of IFX was associated with ATI in 39% of patients versus only 16% in those treated with scheduled maintenance therapy with IFX. These data are highly relevant when considering the debate regarding combination versus monotherapy, because it was clear that administration of a concomitant IS reduced the development of ATI and was associated with higher IFX drug levels in patients receiving episodic IFX ( Table 2 ; also see Table 1 ).

The optimal approach regarding concomitant IS therapy is more uncertain with scheduled maintenance anti-TNFα therapy, not only with IFX but also later with adalimumab and certolizumab pegol. The results of several large randomized controlled trials (RCTs) were undertaken for all three anti-TNFα agents, with approximately 40% of patients entering these respective studies on concomitant immunosuppressive therapy (primarily AZA/6-MP). Retrospective analyses were performed to determine if there was a clinical difference, based on the Crohn Disease activity index (CDAI), between those on combination therapy versus anti-TNFα monotherapy. Lichtenstein and colleagues combined the clinical efficacy rates, infection and serious infection rates, as well as drug pharmacokinetics for all patients in the Accent I and Accent II trials for CD as well as those patients enrolled in the ACT 1 and ACT 2 trials for moderate to severe ulcerative colitis (UC). They found that concomitant administration of IS did not result in any difference in efficacy or exert any change in drug pharmacokinetics. Similarly, retrospective subgroup analyses of the CHARM trial for adalimumab and the Precise-2 trial for certolizumab pegol failed to demonstrate a difference in CDAI-defined response when patients were stratified by baseline use of a concomitant IS agent. However, these studies represent retrospective, post hoc analyses for which the studies under question were not powered. Moreover, they lack the more robust determination of objective end points such as mucosal healing. With respect to pharmacokinetic data for adalimumab and certolizumab pegol, there are less data with respect to the effect of concomitant IS therapy. In the CLASSIC-II study of adalimumab in CD, 3.8% of patients developed antibodies to adalimumab (AAA) with monotherapy versus no patients on combination therapy; however, this result must be interpreted with caution because only 7 of 269 patients developed AAA, making the clinical relevance uncertain. In the Precise-2 trial of certolizumab pegol for CD, there was a numerically higher rate of antidrug antibodies in patients on monotherapy versus those on combination therapy (12% vs 2%, respectively).

To address the question from a different angle, Van Assche and colleagues studied whether patients on combination therapy could safely discontinue the IS agent. They conducted an RCT in which patients in remission on IFX in combination with an IS were randomized to continue their IS agent in combination with IFX or to discontinue IS therapy and continue with IFX monotherapy. There was no statistical difference in the predefined clinical end point (dose change of IFX) at 104 weeks, nor a difference in rates of mucosal healing. However, there was a statistically higher C-reactive protein (CRP) and lower trough IFX levels in those on monotherapy, leading many to speculate that the potential deleterious effects of IS withdrawal may be a problem in a time frame longer than the duration of this study.

At this point in time the data seemed to indicate an apparent lack of benefit from the addition of an IS agent in CD (and UC when considering the ACT data). Then, data emerged that implied that there were higher rates of opportunistic infections with combination therapy, and there was a general acceptance that the use of thiopurines was associated with an increased rate of non-Hodgkin lymphoma (NHL) (statistically significant in younger male patients). In addition, great concern arose around case reports of hepatosplenic T cell lymphoma (HSTCL) in young male patients treated with thiopurines and anti-TNFα therapy. Therefore, there seemed to be a paradigm shift away from combination therapy, particularly among pediatric gastroenterologists. There was a need for further data to clarify these issues in order to determine whether there truly was a benefit with combination therapy and to better ascertain the true significance of HSTCL. The importance of this question was compounded by the fact that, with the exception of natalizumab, at that time practitioners had no other biologic agents at their disposal for the treatment of CD and UC; thus, optimization of therapy was paramount.

No discussion about the issue of combination versus monotherapy can be complete without acknowledging the importance of serum drug levels of anti-TNFα agents. Increasingly, it has become apparent that the absolute serum level of IFX is an important factor in determining response to this therapy. Although less clear from available data, the same is probably true of adalimumab. As discussed later, this factor may be the explanation for the increased efficacy of combination therapy (see Table 2 ).

What are the Infectious Risks Associated With Combination Therapy?

With respect to the risk of infection, it is generally agreed that the use of IS therapy is associated with risk, but quantifying that risk is challenging because there are less controlled data available for IS use than for anti-TNFα therapy. Controlled data come from large RCTs for anti-TNFα agents, in which consistently approximately 40% of patients entered on an IS agent. A recent metaanalysis of all anti-TNFα trials in CD evaluated safety and efficacy. The safety analysis of this study found that anti-TNFα therapy was not associated with an increased risk of infectious complications when compared with CD patients not treated with anti-TNFα. Because there was general consistency in the trials with respect to the proportion of patients on IS in both treatment arms, one could infer that combination therapy is not associated with increased risk. However, none of the individual trials were powered or stratified for such an analysis, and the length of most trials was 1 year or less, thus providing a more limited opportunity to address this question. Similarly, in two recent RCTs designed to explicitly explore the benefit of combination therapy with IFX with an IS (see later discussion), there was no difference in infectious adverse events in patients on combination therapy compared with monotherapy. This finding is in contrast to data from a case-control study from the Mayo Clinic in which the use of combination therapy was associated with a significantly increased risks of opportunistic infection (OI). However, on close inspection of this study, the majority of risk came from the combination of IS and corticosteroids and not the combination of IS and anti-TNFα.

Perhaps the best way to estimate the risk of infection with IS monotherapy versus combination therapy is to look at real-life patients. A study by Marehbian and colleagues provides a good example of this approach, in which they examined claims data from over 22,000 patients in the United States between 2002 and 2005. They calculated the relative rates of adverse events related to having CD, and to medications for CD by comparing with referent patients on no therapy for CD and to over 100,000 control patients without IBD. In this study, two groups of CD patients were studied: (1) prevalent CD patients representing patients with existing CD and (2) a longitudinal cohort representing patients with CD diagnosed within 1 year in an effort to model an inception cohort. Compared with the general population, patients in the prevalent cohort treated with IS agents alone had rate ratios for OI, herpes zoster, and sepsis of 3.63, 2.54, and 1.28, respectively, but these ratios were not statistically significant. When an IS was used in combination with anti-TNFα, the rate ratios were 6.18, 4.15, and 1.75, respectively, and still not statistically significant. The use of combination therapy in the prevalent cohort was, therefore, associated with numerically higher (but not statistically significant) rates of infectious adverse events. It is worth noting, however, that adding corticosteroids was associated with even higher rates when used in combination with other therapies, a phenomenon consistent with other databases. For longitudinal patients treated with an IS and an anti-TNFα agent in combination, only sepsis was significantly elevated with a hazard ratio of 1.73 (confidence interval [CI] 1.17–2.54). Again, when corticosteroids were added to combination therapy, the absolute rates increased and the hazard ratios became significant for candidiasis and sepsis. Taken as a whole, this real-life data representing a very large cohort of patients supports the fact that combination therapy with an IS and an anti-TNFα agent leads to a slightly increased, but not statistically significant, risk of infectious adverse events. Corticosteroids seem to confer the greatest risk, which is perhaps the most important and consistent message practitioners should consider.

What are the Infectious Risks Associated With Combination Therapy?

With respect to the risk of infection, it is generally agreed that the use of IS therapy is associated with risk, but quantifying that risk is challenging because there are less controlled data available for IS use than for anti-TNFα therapy. Controlled data come from large RCTs for anti-TNFα agents, in which consistently approximately 40% of patients entered on an IS agent. A recent metaanalysis of all anti-TNFα trials in CD evaluated safety and efficacy. The safety analysis of this study found that anti-TNFα therapy was not associated with an increased risk of infectious complications when compared with CD patients not treated with anti-TNFα. Because there was general consistency in the trials with respect to the proportion of patients on IS in both treatment arms, one could infer that combination therapy is not associated with increased risk. However, none of the individual trials were powered or stratified for such an analysis, and the length of most trials was 1 year or less, thus providing a more limited opportunity to address this question. Similarly, in two recent RCTs designed to explicitly explore the benefit of combination therapy with IFX with an IS (see later discussion), there was no difference in infectious adverse events in patients on combination therapy compared with monotherapy. This finding is in contrast to data from a case-control study from the Mayo Clinic in which the use of combination therapy was associated with a significantly increased risks of opportunistic infection (OI). However, on close inspection of this study, the majority of risk came from the combination of IS and corticosteroids and not the combination of IS and anti-TNFα.

Perhaps the best way to estimate the risk of infection with IS monotherapy versus combination therapy is to look at real-life patients. A study by Marehbian and colleagues provides a good example of this approach, in which they examined claims data from over 22,000 patients in the United States between 2002 and 2005. They calculated the relative rates of adverse events related to having CD, and to medications for CD by comparing with referent patients on no therapy for CD and to over 100,000 control patients without IBD. In this study, two groups of CD patients were studied: (1) prevalent CD patients representing patients with existing CD and (2) a longitudinal cohort representing patients with CD diagnosed within 1 year in an effort to model an inception cohort. Compared with the general population, patients in the prevalent cohort treated with IS agents alone had rate ratios for OI, herpes zoster, and sepsis of 3.63, 2.54, and 1.28, respectively, but these ratios were not statistically significant. When an IS was used in combination with anti-TNFα, the rate ratios were 6.18, 4.15, and 1.75, respectively, and still not statistically significant. The use of combination therapy in the prevalent cohort was, therefore, associated with numerically higher (but not statistically significant) rates of infectious adverse events. It is worth noting, however, that adding corticosteroids was associated with even higher rates when used in combination with other therapies, a phenomenon consistent with other databases. For longitudinal patients treated with an IS and an anti-TNFα agent in combination, only sepsis was significantly elevated with a hazard ratio of 1.73 (confidence interval [CI] 1.17–2.54). Again, when corticosteroids were added to combination therapy, the absolute rates increased and the hazard ratios became significant for candidiasis and sepsis. Taken as a whole, this real-life data representing a very large cohort of patients supports the fact that combination therapy with an IS and an anti-TNFα agent leads to a slightly increased, but not statistically significant, risk of infectious adverse events. Corticosteroids seem to confer the greatest risk, which is perhaps the most important and consistent message practitioners should consider.

Is There an Increased Risk of NHL With Combination Therapy, and How Should We Approach the Risk of HSTCL?

Before addressing the issue of NHL related to therapy, one must first address whether or not CD itself is associated with an increased risk of NHL. A number of studies have found no association between CD and NHL. One study by Bernstein and colleagues found a statistically significant association in younger male patients, independent of treatment, but this finding has not been replicated by others. There are now clear and relatively consistent data that the use of AZA/6-MP is associated with an increased risk of NHL. A metaanalysis of six studies by Kandiel and colleagues found a pooled relative risk of 4.18 (95% CI 2.07–7.51; 11 observed cases, 2.63 expected). However, this result included patients with both CD and UC, and both NHL and Hodgkin lymphoma. When more closely evaluating these data, there were 4 cases of NHL in 11,012 patient-years of exposure in patients with CD. This result calculates to an absolute risk of 3.6 NHL per 10,000 patient-years. A more recent French cohort study (CESAME cohort) evaluated over 19,000 patients with IBD (approximately 60% CD, 40% UC/IBD unspecified) among whom 23 new cases of lymphoproliferative disease were diagnosed (22 of 23 NHL, 1 of 23 Hodgkin disease). The multivariate-adjusted hazard ratio of lymphoproliferative disease in patients treated with AZA/6-MP was 5.28 (2.01–13.9). There is a paucity of data to draw on with respect to determining if MTX is associated with any increased risk of NHL, but a case-control study in patients with rheumatoid arthritis (RA) failed to demonstrate any increased risk. However, this association represents an area where data are lacking in IBD.

Determining if there is an increased risk of NHL in the context of anti-TNFα therapy is challenging by virtue of the rarity of the event. Ascertaining whether combination therapy is associated with even greater risk of NHL than with IS alone is confounded by the fact that, historically, the majority of patients treated with anti-TNFα therapy had been previously exposed to IS therapy. One of the most comprehensive attempts at addressing this issue came in the form of a metaanalysis of several study types evaluating the risks of NHL in CD patients treated with anti-TNFα therapy. A total of 26 studies met entry criteria, representing almost 9000 patients and more than 20,000 patient-years of anti-TNFα exposure. The crude rate was 6.1 per 10,000 patient-years compared with the expected rate from the SEER (Surveillance Epidemiology and End Results) database, which was 3.23. When stratified by age and gender, the standardized incident ratio (SIR) was significant only in men aged 20 to 54 (SIR 5.4, 95% CI 1.3–18.1). Of the 13 patients with NHL identified in this metaanalysis, 11 had been exposed to IS therapy, thus making a determination of the risk of anti-TNFα alone impossible.

To examine real-life, the study by Marehbian and colleagues provides some insight from a large US sample of CD patients treated with a variety of therapies. In the prevalent CD population, IS monotherapy was associated with a rate ratio of lymphoma (when compared with the general population) of 1.66 (95% CI 0.95–2.88). For anti-TNFα monotherapy, the corresponding rate ratio in this cohort was 2.10 (95% CI 0.82–5.40) and for combination therapy the rate ratio was 0.63 (95% CI 0.08–4.64). None of these comparisons were statistically significant.

Taken cumulatively, it seems that AZA/6-MP is associated with a slightly increased risk of NHL. Whether monotherapy with anti-TNFα is independently associated with an increased risk is uncertain. If combination therapy itself increases the risk of NHL further, the magnitude of the effect is small.

A separate discussion is warranted with respect to HSTCL. This rare but aggressive and almost universally fatal form of NHL was first described in 1990 and is the result of clonal expansion of γδ or αβ T cells. The flurry of case reports beginning in 2006 in the context of patients treated with IFX and azathioprine led to a Food and Drug Administration (FDA) black box warning being added to the IFX package insert later that year. . A recent review of all reports of HSTCL submitted to the FDA found a total of 25 cases, of which 88% were in patients with IBD. Although the majority were younger male patients taking combination therapy, a number of cases on IS monotherapy have been reported, and 16% of cases were in women. Furthermore, 16% of patients were older than 65 years of age, implying that the risk is not isolated to any single age or gender group. It is not possible to calculate an accurate rate of occurrence because the denominator of exposed patients in the highest risk group (young men) is not currently available. However, the absolute rarity of the event can still provide us with some reassurance.

What is the Benefit of Combination Therapy?

As with many issues relevant to the use of anti-TNFα therapy for IBD, the use of the same therapies in rheumatologic disorders such as RA can offer some insight. The question of the benefit of combination therapy in the context of RA is questioned to a significantly lesser degree because it has been repeatedly demonstrated that the concomitant administration of MTX is associated with improved clinical response, but also a reduction in immunogenicity and higher IFX blood levels.

A recent French study evaluated 121 consecutive IBD patients (23 UC, 98 CD) being treated with IFX (for at least 12 months) and evaluated important clinical end points in those receiving concomitant IS therapy with either AZA or MTX (for at least 6 months) versus those on IFX monotherapy. The investigators divided the patient follow-up into semesters, with a semester being a 6-month block of time. They then evaluated exacerbations of IBD, perianal complications, dose escalation, and switch to adalimumab by semesters with and without IS therapy. Overall, semesters with combination therapy were associated with fewer disease exacerbations, perianal complications, and a switch to adalimumab than those with monotherapy (19.3% vs 32% P = .003, 4.1% vs 11.8% P = .03, 1.1% vs 5.3%, P = .006).

The best way to properly address the question surrounding the benefit of combination versus monotherapy in IBD is through prospective RCTs designed specifically to address this issue. Fortunately, we now have four such trials from which we can draw some conclusions.

The first trial to examine this question evaluated whether an IS could safely be withdrawn when a patient was already on and responding well to combination therapy with IFX. As noted previously, there was no clinical difference at 2 years in terms of a need for a change in IFX dosing or in mucosal healing, but there was a difference in CRP and IFX pharmacokinetics that favored combination therapy.

The largest and most definitive attempt at examining this question of combination versus monotherapy is the Study of Biologic and Immunomodulator Naïve Patients in Crohn Disease (SONIC). The SONIC trial randomized 508 patients with active, moderate to severe CD who were naïve to IS and biologic therapy to either AZA alone (n = 170) at a dose of 2.5 mg/kg, IFX induction and maintenance alone (n = 169), or combination therapy with AZA and IFX together (n = 169). The primary end point was corticosteroid-free remission at 26 weeks, but important secondary end points included mucosal healing at 26 weeks (among patients with ulceration noted at baseline), corticosteroid-free remission at week 50, as well as IFX pharmacokinetics and determination of ATI. The primary end point was achieved in the AZA alone group in 30%, with IFX monotherapy in 44% and in 57% of patients on combination therapy (both IFX groups were statistically significantly superior to AZA alone ( P = .006 IFX vs AZA, P <.001 IFX/AZA vs AZA). Strikingly, mucosal healing was noted in only 16.5% of patients with AZA alone, versus 30% and 43.9% of patients treated with IFX monotherapy and combination therapy, respectively ( P = .02 IFX vs AZA, P <.001 IFX/AZA vs AZA). ATI were noted in 14.6% of patients treated with IFX monotherapy versus 0.9% of patients treated with combination therapy. Moreover, median trough IFX levels in patients treated with monotherapy were 1.6 μg/mL versus 3.5 μg/mL in patients in the combination therapy group, and rates of corticosteroid-free remission were higher in patients with higher trough IFX levels.

The SONIC trial provides the most resounding evidence that combination therapy (in this case with a thiopurine) is superior to monotherapy with IFX alone, and certainly superior to AZA monotherapy. The rate of infections was nearly equivalent across the three groups, and fortunately, there were no cases of lymphoma. However, SONIC was still too small to understand how rare adverse events might impact the benefit-risk tradeoff with combination therapy. To address this issue, a recent decision analysis studied how often rare adverse events (such as sepsis or lymphoma) would have to occur with combination therapy to change the overall benefit in favor of IFX monotherapy. At baseline, using SONIC as a model, over a 1-year period, AZA/IFX combination therapy is the favored treatment strategy. For the model results to change (ie, the risks of combination therapy to outweigh the benefits), the rate of serious infections on combination therapy would have to be 20% and the rate of NHL 3.9%. These rates are 5 and 65 times higher than what has been seen in the literature and seem higher than what is clinically realistic. Therefore, the conclusions were that in most cases, even when considering rare adverse events, the benefits of combination therapy outweigh its risks.

However, it is important to recognize a few important caveats in terms of the generalizability of SONIC. First and foremost, all patients were naïve to IS and anti-TNFα. Additionally, patients had a shorter disease duration (<3 years) than most patients having entered prior RCTs for anti-TNFα agents. Therefore, these data may or may not apply equally to a patient with a longer disease duration who starts an anti-TNFα agent after failure of AZA/6-MP (a typical example of a patient entering prior anti-TNFα trial).

The third trial to evaluate the issue of combination therapy is a multicenter RCT evaluating IFX induction and maintenance with or without subcutaneous MTX for patients with moderate to severely active CD (COMMIT trial). In this Canadian study, patients with moderate to severely active CD were treated with a per-protocol corticosteroid induction and taper in conjunction with either IFX induction and maintenance (n = 63) or IFX in combination with 25 mg MTX subcutaneously weekly (n = 63). The primary end point was time to treatment failure, defined as failure to achieve corticosteroid-free remission at week 14 or maintain corticosteroid-free remission out to week 50. At week 50, 30.6% of patients in the MTX/IFX group had failure of treatment versus 29.8% of those on IFX alone ( P = .63). Pharmacokinetic data from COMMIT demonstrated that 4% of patients on combination therapy developed ATI versus 20.4% of patients on IFX monotherapy ( P = .01). In patients treated with MTX, the median trough IFX level was 6.35 verus 3.75 in those on monotherapy, and patients with detectable trough IFX were numerically more likely to be a treatment success (72%) than those with no detectable trough IFX (52%). None of the pharmacokinetic comparisons were statistically significant.

There are important similarities and differences between the SONIC and COMMIT trials that are relevant to combination versus monotherapy debate. On the surface, the results seem discordant and could potentially leave the practitioner with more questions than answers. However, the study populations and the study designs warrant further discussion. As noted previously, the SONIC population represents patients with a shorter disease duration, many of whom were requiring some amount of corticosteroids, but there was no per-protocol corticosteroid induction and taper. In COMMIT, all patients received a full corticosteroid induction, so this trial is really a comparison of triple induction and dual maintenance versus double induction and single-therapy maintenance. It is important to reflect on the very high rates of corticosteroid-free remission in both treatment arms in COMMIT, rates that are far in excess of those reported in adult anti-TNFα trials previously. This result has led many to conclude that corticosteroid induction may further optimize response and that perhaps this response could serve as an explanation as to why no clinical difference was noted between the two groups (eg, a “ceiling effect”). Another key difference is the absence of mucosal healing data in the COMMIT trial, so one can only speculate as to whether there would have been a difference in this end point. Perhaps the most important similarity to SONIC that proponents of combination therapy will point out is the similarity in the immunogenicity and pharmacokinetic data between the two trials, both favoring combination therapy. Whether AZA is a superior combination therapy agent to MTX cannot be concluded from these trials because of the important differences noted previously. As seen in SONIC, there was no increase in the rate of infections in patients treated with combination as compared with IFX monotherapy.

The fourth and most recent trial to evaluate this question was a multicenter RCT evaluating different treatment strategies in 231 patients with mild to moderately active UC for whom corticosteroids were failing (UC-SUCCESS). Patients had to be naïve to AZA (or had stopped AZA ≥3 months prior to entry) and anti-TNFα. Patients were randomized to AZA, IFX induction and maintenance, or the combination of both. The primary end point was corticosteroid-free remission at week 16 with secondary end points including response and mucosal healing. The primary end point was met in 22% of patients in the AZA treatment group, 24% in the IFX group, and 40% in the combination therapy group ( P <.05 compared with IFX and AZA groups). There was a stepwise increase in mucosal healing rates at week 16 in the three groups at 37% with AZA monotherapy, 55% with IFX monotherapy, and 63% with combination therapy ( P <.05 for both IFX groups compared with AZA). No pharmacokinetic data are available currently from this study. Despite the discordantly low rate of clinical remission with IFX monotherapy at 16 weeks in this study, the findings as a whole seem to favor combination therapy.