Main findings
The OPTION-DM trial showed that all three treatment pathways (i.e. A-P, D-P and P-A) were equally effective in reducing neuropathic pain intensity, significantly reducing the baseline NRS score from a baseline mean of 6.6 (SD 1.5) points to an average of 3.3 (SD 1.8) points at 16 weeks. There were no differences between the treatment pathways, as the effects observed were not different statistically or clinically different between the pathways, with the CIs for the primary end point (i.e. pain at week 16) having the prespecified difference only at their very edge.
As the three treatment options produced similar pain control, it might be important to consider other outcomes when deciding which treatment to recommend; however, no one treatment was uniformly preferable across the end points we assessed.
The P-A pathway led to the fewest discontinuations on monotherapy due to AEs (p = 0.031) and numerically was the most preferred pathway (A-P, 24%; D-P, 33%; P-A, 43%; p = 0.27) at the end of the trial. This is consistent with the American Academy of Neurology recommendations, which give only pregabalin level A evidence, mainly because completion rates in pregabalin trials were more than 80%.79 The P-A pathway also gave the best pain relief for higher levels of baseline depression (p = 0.011) and both the P-A and D-P pathways provided better pain relief than the A-P pathway for higher levels of baseline anxiety (p = 0.016).
The trial explored a number of secondary end points. The head-to-head trial of maximum tolerated doses of amitriptyline, duloxetine and pregabalin showed similar efficacies for all three monotherapies at the end of 6 weeks, although there were fewer discontinuations due to treatment-emergent AEs with pregabalin. However, only in one-third of patients did monotherapy achieve mild pain (i.e. ‘responders’, with a NRS score of ≤ 3). Similarly, monotherapy achieved 50% pain relief in around 40% of patients. Most patients (‘non-responders’) required combination treatment, which resulted in a 0.6-point further improvement in NRS score and 18% and 14% increases in the achievement of a NRS score of ≤ 3 points and 50% pain relief, respectively. In a crossover trial, Gilron et al.57 studied 56 patients with neuropathic pain (40 patients with DPNP) treated to maximum tolerated doses of gabapentin, nortriptyline and their combination over 1-month treatment periods. They found that combination treatment was more efficacious than either drug alone. Another crossover trial80 (n = 73), of 5-week treatment periods, compared the combination of imipramine and pregabalin at moderate doses with either treatment on its own in painful polyneuropathy. The study found that combination treatment was more efficacious in relieving neuropathic pain than either drug on its own, but resulted in higher rates of side effects. Despite these results, currently, a number of international bodies do not recommend combination treatment for DPNP because of ‘insufficient evidence’.81–83 Although the OPTION-DM trial was not designed as a comparison of monotherapy and combination treatment, the data make a compelling case for the recommendation of combination treatment of first-line drugs for DPNP patients with suboptimal response to a monotherapy.
Despite massive variations in the cost and availability of amitriptyline, duloxetine and pregabalin in many countries, it is reassuring that all three drugs are equally efficacious in relieving pain. Furthermore, all monotherapies resulted in a substantial improvement of QoL domains, sleep and measures of mood from baseline. However, amitriptyline was significantly better than duloxetine in improving physical functioning and sleep, and pregabalin was superior to duloxetine in improving role limitation due to physical health. Treatment-emergent AEs were predictable for the monotherapies, although there were no significant differences in their overall frequency and severity. This may, in part, be due to the use of maximum tolerated doses rather than a fixed treatment regime. There were also no significant differences in the frequency of SAEs. However, compared with some previous studies, combination treatment with maximum tolerated doses was well tolerated, with few treatment-emergent AEs, except for larger numbers of patients reporting diarrhoea and dry mouth in the A-P pathway.
Health economics
To the best of our knowledge, this study is the first to present the results of a head-to-head comparison of current medications and their combinations in an economic evaluation over a 16-week period. QALYs were similar over the study time period, with an average difference in costs ranging from –£113 to £154 across the three comparisons. Ninety-five per cent bootstrap CIs around costs and QALYs demonstrated the uncertainty in results. Furthermore, the study showed that greater costs are borne by the patients and their carers rather than the NHS.
Wu et al.32 examined the cost-effectiveness of duloxetine compared with usual care over a 52-week period. Wu et al.32 looked at three perspectives and found incremental costs slightly higher than those for our study [i.e. direct (closest possible to NHS) US$1600 (£1194); employment and direct US$2196 (£1639) and societal, direct and employment US$2754 (£2056)]. However, in addition to looking at single rather than combination therapy, the time period is longer in the Wu et al.32 study and the health-care system is different in the USA and, therefore, caution should be given when comparing these results with those from the OPTION-DM trial.
Two further studies33,34 looked at duloxetine in decision-analytic models. O’Connor et al.34 presents the cost per QALY of duloxetine compared with desipramine at US$47,700 (or £35,609). This is higher than our average QALY for A-P compared with D-P (£7026) and lower than the average QALY for D-P compared with P-A (£94,157), although it should be noted, again, that this is not a combination therapy study and the time period is shorter and so not comparable. It should also be noted that the O’Connor et al.34 results were sensitive to obtaining pain relief. Beard et al.33 also fitted a decision-analytic model and noted a saving of £77,071 for every 1000 patients treated in 2007 or £109,246 per 1000 patients in 2019.
Finally, de Salas-Cansado et al.35 compared pregabalin with usual care from a Spanish NHS and societal perspective over a 23-week period. The costs per QALY were higher than those for our study [i.e. €5302 (£4426) societal and €14,381 Spanish NHS (£12,005)].
Also worth noting is that the utility values for our sample do appear to be low in comparison with those for the general diabetes population. Table 3 presented mean utility values for the EQ-5D at baseline and at 6 and 16 weeks, and these ranged between 0.408 and 0.551. Collado Mateo et al.84 present normative values for a Spanish diabetes population by age and sex. The mean EQ-5D values for people aged 60–69 years (the most comparable age band to our population mean of 61.8 years) were 0.860 for males and 0.745 for females, respectively. The authors used the Spanish crosswalk rather than UK crosswalk to obtain utilities for the EQ-5D, but other studies also suggest higher utility values. For example, Wang et al.85 report a mean utility of 0.720 for patients in Singapore with type 2 diabetes mellitus. Likewise, in a Chinese study, Pan et al.86 report a mean value of 0.862 for patients with type 2 diabetes mellitus and a mean of 0.815 if the patients also had neuropathy.
Strengths and weaknesses of the research
To the best of our knowledge, this was the first randomised double-blind comparator trial of neuropathic pain treatment pathways. Although head-to-head trials of individual monotherapies and combination treatments could be designed, investigating treatment pathways as a whole was felt most efficient and applicable to current clinical practice. This is because most patients are started on a monotherapy and will require a second agent to be added in combination within a few months. Therefore, the OPTION-DM trial reflected current practice, allowing the outcomes of this study to be readily generalisable. Recruitment from a wide array of sites, including primary and secondary care, further strengthens the generalisability of the study.
The initial prespecified number of participants was not recruited, and this could be taken as a limitation of the study. Early monitoring of the fidelity of recruitment revealed that recruitment for this demanding trial was challenging. Rather than compromise the integrity of the trial, and with the approval of the independent TSC and head of the Health Technology Assessment, the trial continued to a fixed time point when an adequate sample size had been achieved to detect a difference of at least 1 NRS point. The onset of the COVID-19 pandemic further affected follow-up, with some participants having a curtailed follow-up in the final pathway. A strength of the OPTION-DM trial was that, although the trial was curtailed in terms of recruitment, to the best of our knowledge, it is still the longest and largest crossover combination trial. Given the results, and that a 0.5-point NRS difference was at the very edge of plausible difference between the treatment arms, even a significantly larger sample size would probably not have altered study outcomes or conclusions.
The lack of a placebo arm can also be taken as a limitation of the study. However, given the existing wealth of evidence from previous placebo-controlled trials and the advice from our PPI panel, we decided against a placebo control, as including a placebo control would have made the trial longer, more complicated (i.e. a four-way crossover trial) and it would not have been ethically justifiable to leave participants on placebo alone for 17 weeks. The number of dropouts within the 1–2 weeks prior to randomisation in this trial suggest that such a trial is not feasible. The lack of a placebo arm means that we cannot attribute the change from baseline solely to the introduction of treatment, as some improvement may be due to regression to the mean. Nonetheless, previous trials14–22 have demonstrated far smaller changes in participants treated with placebo, and it is reasonable to assume that most of the improvement is due to therapy.
The EQ-5D was captured prior to treatment pathway 1, but not prior to treatment pathways 2 and 3, and was captured for each treatment pathway at 6 and 16 weeks. It may be that measuring HRQoL more frequently would more likely capture any disutilities associated with AEs, and any future study should consider this. The economic analysis reports the within-trial results over 16 weeks, and this is in line with other published economic evaluations,31,32 although future studies could consider modelling the cost-effectiveness over a longer time frame.
There are two further limitations of the economic evaluation. First, the use of a questionnaire to capture resource use means that resource use is self-reported and could be subject to recall issues, thereby potentially underestimating the overall cost of each treatment pathway. The capturing of resource use information from routine sources could be considered in future studies. Second, the crossover nature of this study lent itself to a pairwise comparison of the alternative treatment pathways. This pairwise comparison was varied in a sensitivity analysis, which, owing to the small difference in costs and QALY, meant that the results remained uncertain.
Evidence in the context of other existing research
The OPTION-DM trial was an efficiently designed head-to-head crossover trial,2 with each patient undergoing all three pathways over 50 weeks, making it, to the best of our knowledge, the longest blinded neuropathic pain trial. The durations of monotherapy and combination treatment were long enough to assess full treatment effects.37 Previous combination trials mainly used fixed-dose titration regimens regardless of treatment response.31,80 This does not reflect clinical practice and often resulted in high dropout rates.31 The present trial employed a flexible dosing regimen to achieve maximum tolerated doses, based on individual responses. Moreover, all previous DPNP multiperiod crossover trials were smaller.25–27 In addition, other studies included neuropathic pain conditions other than DPNP.29,57,80 Finally, most previous trials did not use combinations of current first-line drugs.29,57,80 The OPTION-DM trial tried to obtain robust data from current first-line drugs, as, based on the trajectory of new drug developments for DPNP over the past 25 years, the emergence and use of revolutionary new drugs that are considerably more efficacious than current ones seemed unlikely over the foreseeable future.
Implications for practice and policy
The OPTION-DM trial has demonstrated that all three treatment pathways and monotherapies are equally efficacious. Although most international painful diabetic neuropathy management guidelines, including the one by NICE (CG96, updated in 2021),82 place the monotherapies as first-line agents, based on indirect comparisons of the treatments’ efficacy, this is, to the best of our knowledge, the first time that an adequately designed head-to-head comparator trial has been undertaken, and it confirmed that the treatments are of equivalent efficacy. The OPTION-DM trial will hopefully reduce the ambiguity of indirect comparisons that, for example, led the French panel of experts to recommend pregabalin as a second-line agent.87
Combination treatment is currently not recommended by international guidelines because of insufficient evidence.81–83 NICE guidance,82 which was updated in 2021, recommends switching to a different treatment rather than combination treatment. In this trial, only one-third of patients saw a reduction in pain to ‘mild’ as quantified as a NRS score of ≤ 3 points (‘responders’). Similarly, monotherapy achieved 50% pain relief in around 40% of patients. Many patients (‘non-responders’) required combination treatment that resulted in a further mean improvement of 1 point on the NRS and an additional 18% and 14% increase in the achievement of NRS score of ≤ 3 points and 50% pain relief, respectively. Although the OPTION-DM trial was not designed as a comparison of monotherapy and combination treatment, the data make a compelling case for the recommendation of combination treatment of first-line drugs for DPNP patients with suboptimal response to a monotherapy, particularly as combination treatment was very well tolerated.
In clinical practice, patients are sometimes prescribed very high doses of amitriptyline (up to 150 mg/day) and, therefore, it is no great wonder that so many patients experience unwanted side effects. The OPTION-DM trial has shown that titration to 60 mg per day can be effective. The OPTION-DM trial also found that the mean dose per day (% on maximum dose) of amitriptyline, duloxetine and pregabalin at week 6 was 60 mg (59%), 82 mg (53%) and 396 mg (56%), respectively. For patients on combination treatments A-P, D-P and P-A the mean dose per day (% on maximum dose) of add-on drug at week 16 was 365 mg (47%), 407 mg (55%) and 55 mg (47%), respectively. These data show that only around half of patients will tolerate maximum doses of these drugs, and this is in the context of a clinical trial that excluded patients with major comorbidities. A key message from the OPTION-DM trial is that drugs need to be titrated gradually, over at least 2 weeks and, in some patients (for example the elderly and those with comorbidities) even more slowly, factoring in the side effect profile and the severity of side effects, as well as efficacy.
The data from OPTION-DM trial are powerful. The OPTION-DM trial was, to the best of our knowledge, the longest blinded neuropathic pain trial to date. Unlike previous combination treatment crossover trials,37 the durations of monotherapy and combination treatments were long enough to assess full treatment effects. Moreover, previous combination trials31,80 mainly used fixed-dose titration regimens regardless of treatment response, which does not reflect clinical practice and resulted in high dropout rates. The OPTION-DM trial employed a flexible dosing regimen to achieve maximum tolerated doses, based on individual responses. Moreover, all previous DPNP multiperiod crossover trials25–27 had smaller sample size and other studies29,57,80 included neuropathic pain conditions other than DPNP. In addition, most previous trials did not use combinations of current first-line drugs. Therefore, the robustness of OPTION-DM trial data will probably influence clinical practice, guidelines and commissioners.
Implications for health care
The current NICE guideline (CG173)82 for the management of painful DPNP states:
Offer a choice of amitriptyline, duloxetine, gabapentin or pregabalin as initial treatment for neuropathic pain.
If initial treatment is not effective or is not tolerated, offer one of the remaining 3 drugs, and consider switching again if the second and third drugs tried are not effective or not tolerated.
The OPTION-DM trial was a pragmatic trial reflecting current best-treatment practice and examined three treatment pathways for painful diabetic neuropathy. In each treatment pathway, patients were initially commenced on monotherapy followed by combination therapy. The results showed significant improvements in pain severity in all three treatment pathways. In addition, the results showed that combination treatment is not only safe but also necessary for a significant proportion of patients to achieve meaningful pain relief (i.e. a NRS pain score of ≤ 3 points).
The most recent NICE CG173 guideline82 recommends that patients switch to a second or third agent if initial/subsequent monotherapy is ineffective. In practice, this often results in poor patient satisfaction and disillusionment, as well as failure to achieve adequate pain relief due to treatment inertia. The results of the OPTION-DM trial suggest that combination treatment may be advocated if inadequate pain relief is achieved with monotherapy. To facilitate this, future pharmacotherapy guidelines for painful diabetic neuropathy should be rationalised into treatment pathways, as the majority of patients end up having combination treatments. Our cost-effectiveness analysis shows that there are no differences between the treatment pathways assessed. Further post hoc analyses are being planned to examine if certain subgroups of patients (e.g. the elderly, patients with mood disorders or clinical pain phenotypes) respond better to a particular treatment pathway.
Recommendations for future research
Implementation
There remains a clinical and economic need to improve outcomes in patients with painful diabetic neuropathy. Based on the trajectory of new drug developments for painful diabetic neuropathy over the past 25 years, the emergence and use of new drugs that are more efficacious than current ones seems unlikely in the next decade. In the OPTION-DM trial, around 40% of patients achieved significant improvement in pain (i.e. 30% improvement in NRS or a NRS score of ≤ 3 points) based on simple pragmatic guidance to treatment escalation. There is, however, a wide variation in treatment outcomes across England. We need to explore the differences in clinical practice and patient behaviour that underlie these differences, as well as promote the outcomes of the OPTION-DM trial.
Mechanistic approach
In all treatment pathways, there was a wide variation in NRS pain intensity scores, with some patients showing marked improvement and others showing minimal or no improvement. Further research is needed to explain why some patients do so well, whereas others do not. This research would involve detailed clinical phenotyping of patients using recent innovations in magnetic resonance neuroimaging, quantitative sensory assessment and genotyping. This research might uncover underlying pain mechanisms that inform treatment responses in individual patients.
Holistic approach
It is notable that we found a significant relationship between emotional distress (e.g. anxiety and depression) and improvement in pain scores in two of the three treatment pathways assessed. The precise role of psychosocial factors in explaining pain severity, distress and treatment response are not well understood. Most studies have investigated psychosocial and biomedical variables independently and the differing contributions of these variables on treatment outcomes are poorly specified.
Pharmacological and non-pharmacological approaches
A significant proportion (around 50%) of patients are not helped by any treatment pathway and this raises the question of ‘what next?’. NICE recommends tramadol only if rescue therapy is needed, but often patients are commenced and remain on long-term opioid therapy. The safe use of opioids and other pharmacological [e.g. high-dose (8%) capsaicin patch or intravenous lidocaine therapy] and non-pharmacological (e.g. spinal cord stimulation) treatments for painful diabetic neuropathy should be explored further.