Antidepressant medication use for treatment of chronic ocular pain

Introduction

Chronic ocular pain (COP) is a leading cause of eye care visits in the US. Approximately 5–15% of people will experience COP over their lifetime at an estimated cost of over $55 billion annually.^{1, 2} People with COP suffer symptoms of eye burning, aching, and irritation, among others, leading them to seek medical care. While many conditions underlie COP, such as dry eye disease or post-herpetic neuralgia, some people experience pain without significant ocular signs on examination, and without known risk factors such as trauma, surgery, or infection. Because of the often relentless nature of the pain, the impact of COP on patients is staggering including decreased quality of life similar to those with an immobilizing hip fracture or moderate/severe angina.³ COP also decreases visual functioning and is often accompanied by depression, anxiety, and even suicidal ideation.^{4, 5}

Neuromodulators, including antidepressants such as tricyclic antidepressants (TCAs) and serotonin and norepinephrine reuptake inhibitors (SNRIs), are often used to treat patients with COP, particularly after conventional topical therapies and interventions fail to provide pain relief. However, a recent Cochrane review and network meta-analysis investigating 25 different antidepressants concluded that only duloxetine and perhaps milnacipran (both SNRIs), have sufficient data to support their use in chronic pain.⁶ For all other medications, the evidence was of low certainty, leaving eye care providers wondering if antidepressants should be abandoned for the treatment of COP. We aim to answer this question. We pose that while antidepressants are not a panacea for chronic pain or COP, ophthalmologists and optometrists that evaluate and manage patients with COP should understand when and why these medications are helpful as part of a multimodal targeted treatment plan.

What is our current understanding of chronic pain mechanisms?

The International Association of the Study of Pain’s (IASP) overall definition of pain is “an unpleasant sensory and emotional experience associated with, or resembling that associated with, actual or potential tissue damage.”⁷ However, all pain is not the same. Nociceptive pain is caused by damage or trauma to tissues as often seen in corneal ulcers or ocular inflammation associated with dry eye disease. By contrast, neuropathic pain is caused by a lesion in the somatosensory system, including the peripheral and central nervous system, and is believed to be the primary mechanism at play among patients with conditions, such as herpetic keratitis, post-herpetic neuralgia, and post-LASIK ocular pain.

Perhaps less familiar to many clinicians is the concept of nociplastic pain. This third mechanistic class of pain was defined by the IASP in 2017. It has been clearly established that this form of pain arises from altered pain perception and pain processing in the central nervous system without tissue damage that would activate nociception.⁸ Nociplastic pain is characterized by pain out of proportion to objective signs on clinical examination or testing (i.e. sign/symptom discordance), multisite pain, presence of other chronic pain conditions (so called chronic overlapping pain conditions [COPC], e.g., fibromyalgia, irritable bowel syndrome, temporomandibular disorder, and interstitial cystitis/bladder pain syndrome), associated non-pain CNS-mediated symptoms (e.g., fatigue, sleep disturbances, mood disorders, cognitive and memory impairments), and hyper-responsiveness to both painful and innocuous stimuli.⁸

Nociplastic pain has yet to be rigorously defined in COP. But evidence exists that it is present in many individuals. In several studies, COP has been found to co-aggregate in individuals with other COPCs.⁹ As in other chronic pain states, examination findings (e.g. corneal or conjunctival staining) and pain can be discordant.¹⁰ By contrast, correlations between COP and non-ocular pain, depression, and anxiety are more robust.¹¹ Furthermore, many patients with COP have other chronic CNS-mediated somatic symptoms, such as sleep disorders.¹¹ Neurobiological data from experimental sensory testing has demonstrated that some people with COP experience increased pain sensitivity at remote sites and prolonged aftersensations suggesting decreased descending pain inhibition.¹⁰ In the few small neuroimaging studies of patients with COP, participants had enhanced connectivity in brain regions important for pain processing similar to other nociplastic pain conditions.¹²

Although nociplastic, nociceptive, and neuropathic pain can occur in isolation, many patients have a heterogenous mixed-pain state.⁸ In fact, chronic pain conditions traditionally classified as nociceptive and/or neuropathic (e.g., chronic low back pain, rheumatic disorders, and small fiber neuropathy), are now known to have a nociplastic component as the primary pain mechanism for maintaining chronic pain. For many patients with COP, more than one pain mechanism is present and manifest in varying degrees. Adequate pain management for patients suffering from COP requires addressing our treatments to each of the individual mechanisms at play.

How do antidepressant medications work to alleviate chronic pain?

Instead of using the term “antidepressant” (as do Birkinshaw et al.),⁶ it is preferable to specify the drug class (e.g., SNRI, TCA, selective serotonin reuptake inhibitors [SSRIs]) when considering the analgesic effects of these medications. The term “antidepressant” suggests that all antidepressants act as analgesics or that these drugs are improving pain by improving depression. However, evidence points to a heterogenous effect of antidepressants across drug classes and a pain benefit that is independent of the effect on depression.

Within the various classes of antidepressant drugs, those that inhibit reuptake of norepinephrine (e.g. SNRIs, TCAs, and norepinephrine reuptake inhibitors [NRIs]) are generally thought to be more effective analgesics than those classes that only raise serotonin.¹³ Norepinephrine and serotonin appear to play important roles in decreasing pain signaling in descending pain pathways.¹⁴ TCAs and SNRIs may reduce pain through anti-inflammatory effects by reducing inflammatory cytokines TNF-alpha and IL-6.¹⁵ In contrast, monoamine oxidase inhibitors (e.g., moclobemide) and drugs that have more prominent effects on dopamine (e.g., bupropion) are not effective analgesics. Furthermore, virtually all studies of specific SNRIs, TCAs and NRIs (e.g., duloxetine) suggest that these drugs are not relieving pain solely or largely by acting as an antidepressant. Instead, individuals with chronic pain without depression are nearly as likely to respond to these medications as individuals with depression.¹⁶

How effective are specific drug classes at relieving chronic pain?

Because of numerous clinical trials demonstrating efficacy in reducing pain, TCA, SNRIs, and NRIs are among the most prescribed medications for chronic pain, including COP. The Cochrane review and network meta-analysis by Birkinshaw et al.⁶ sought to determine if specific drug classes, including SSRIs, SNRIs, and TCAs, are more efficacious, and safe across all pain conditions, including fibromyalgia, musculoskeletal pain, gastrointestinal pain, and non-cardiac chest pain. Clinical trials in patients with headache were excluded and none of the clinical trials included examined the impact of these classes of drugs among individuals with COP. Drugs were compared to either placebo, a different antidepressant, a different type of treatment, or a different dose of the same antidepressant. Three primary efficacy outcomes were assessed: substantial pain relief (at least 50% reduction in pain intensity measured using various scales including the Visual Analog Scale and Brief Pain Inventory), pain intensity, and improvement in mood (measured by various scales). Secondary outcomes were moderate pain relief (at least 30% reduction in pain intensity), physical function, sleep, quality of life, and patient global impression of change (these were less frequently reported among the different clinical trials). Only four drugs included data from 10 or more trials: duloxetine (43 studies; 11, 608 participants), milnacipran (11 studies; 5083 participants), amitriptyline (43 studies; 3372 participants), and fluoxetine (11 studies; 630 participants). The study ultimately concluded that there was limited effectiveness of antidepressants on chronic pain across all outcomes, except for two SNRIs: (1) duloxetine for pain relief and physical function, and (2) milnacipran for pain relief to a lesser degree. Conclusions about other drugs could not be made because they lacked enough high quality studies. The average length of the included randomized controlled trials was 10 weeks, so long term efficacy could not be determined. Additionally, assessment of safety across the studies was poor and could not be addressed.

Deciding that all medications, except perhaps duloxetine and milnacipran, are worthless for the management of chronic pain fails to consider the goal of the study and methodologic limitations. The goal was to determine if specific medications were beneficial across chronic pain conditions. Within that context, the beneficial analgesic effect of SNRIs is not surprising. The literature supports that SNRIs, especially duloxetine, are most effective among patients with nociplastic pain. In both osteoarthritis and chronic low back pain, known mixed pain conditions, patients with more widespread pain as measured by Widespread Pain Index (a surrogate measure of nociplastic pain) were more likely to respond to duloxetine than those with regional pain.^{17, 18} Just like osteoarthritis and chronic low back pain, many of the chronic pain conditions included in the Birkinshaw et al.⁶ study were primary nociplastic (i.e., fibromyalgia) or mixed pain conditions where nociplastic pain is known to play a role in maintaining chronic pain. However, even among patients with fibromyalgia, the prototypical nociplastic pain condition, the treatment effects of TCAs and SNRI’s on pain are statistically significant but small when compared to placebo.¹⁹ Thus, it seems SNRIs, such as duloxetine, may be best suited to patients with some degree of nociplastic pain, but are unlikely to provide sufficient pain relief for most patients.

Additionally, the methodology used does not allow us to draw conclusions about whether a specific medication is most effective for a specific diagnosis, let alone an individual patient. As mentioned, none of the studies included COP and treatment effects were pooled across chronic pain conditions in which the underlying mechanisms may vary substantially. Even within the same condition, the effects of any given treatment can vary considerably across patients. This phenomenon is known as heterogeneity of treatment effect. Even if a treatment has no effect in the average study population, often some participants will receive a benefit from the treatment while others may receive less or no benefit. In fact, most of the treatment benefit demonstrated in clinical trials are concentrated in a small group of participants. Therefore, identifying those individuals most likely to benefit from a particular treatment using multivariable risk stratified analyses or other analytic techniques provides nuance to findings.²⁰ These types of targeted analyses are even more challenging in metanalyses because they are resource intensive, but evaluating heterogeneity of treatment effect allows for pooled estimates of average effects across studies and for estimates of how the treatment effect is modified by individual-level and study-level characteristics. A recent Cochrane review compared pooled analyses to individual-level analyses (N=39 studies with 190 comparisons) and found that some conclusions of the meta-analyses would differ based on whether pooled or individual-level analyses were used.²¹ Farrar et al.²² recently examined this issue for interstitial cystitis/bladder pain syndrome, where the National Institutes of Health had funded a series of large randomized controlled trials of a variety of therapies that were nearly all viewed as negative trials. They observed that the subset of individuals with localized, peripherally mediated pain were more likely to respond to peripherally directed therapies, whereas those with more nociplastic pain were more likely to respond to TCAs. As we generate more evidence for appropriate COP treatment algorithms, individual level analyses will be key to helping us detect which patients are most likely to benefit from specific treatments.

How should antidepressant medications be used to manage chronic pain and how does this impact our treatment of COP?

Single treatments for chronic pain, including antidepressants, are often ineffective. Close to 90% of patients with chronic pain have incomplete pain relief with practices that emphasize single therapeutic solutions and trial-and-error treatment approaches.²³ Given the heterogeneity in pain mechanisms at the individual level, multimodal pharmacologic and non-pharmacologic pain interventions, tailored to the individual patient, have become the standard in chronic pain management (Figure 1). The US Department of Health and Human Service recommends multimodal therapies in the Pain Management Best Practices report, but clinical trials of multimodal therapies have not been conducted among patients with COP.²⁴

Figure 1.

Examples of multimodal treatment approaches for ocular pain.

TENS, transcutaneous electrical nerve stimulation; MBSR, mindfulness-based stress reduction; NSAID, non-steroidal anti-inflammatory drug.

Antidepressant medications, specifically those that raise norepinephrine, are one part of multimodal pain interventions. However, non-pharmacologic integrative therapies, such as education, lifestyle changes, and psychotherapeutic interventions (e.g., cognitive behavioral therapies), are the cornerstone of pain management, in particular among patients with nociplastic pain.⁸ Importantly, reductions in pain intensity of 50% are associated with improved self-reports of associated non-pain CNS-mediated symptoms including sleep disturbances, fatigue and depression. These findings suggest that, by appropriately treating pain, clinicians impact other aspect of physical and psychological functioning.²⁵ Multiple clinical trials are underway assessing which patients benefit from different integrative chronic pain treatments and how those treatments work (e.g., NCT04870957, NCT04486066). These treatments include mindfulness-based stress reduction, physical therapy and exercise, duloxetine, and psychotherapeutic interventions.

It is unknown which therapies optimally treat COP. COP itself is a heterogenous, mixed pain condition that can arise from primary or ongoing nociceptive injury (e.g., surgery, infection) or as part of a broader systemic pain condition (e.g., fibromyalgia, migraine, traumatic brain injury). Just as the underlying mechanisms differ, so do the phenotypic presentations (e.g., different characterization of pain, different nerve morphology on in vivo corneal confocal microscopy, different responses to proparacaine challenge). As in other chronic pain conditions, treatment response will vary. Future research will need to validate clinical tools that allow clinicians to rapidly phenotype patients at the point of care. A number of NIH-funded pain networks have used this strategy for other chronic pain conditions such as chronic low back pain. These data would support clinical trials of both pharmacologic and non-pharmacologic interventions for COP targeting specific pain mechanisms.

Conclusion

Patients with COP and treatments for COP are not a monolith. Meta-analyses are important tools for pooling treatment effects across studies, but the findings must be considered in context of the study goals and limitations. It is premature to conclude that antidepressants have no place in the treatment of COP. Clinicians need dynamic clinical tools to rapidly identify clinically relevant COP subtypes, to match subtypes to treatments, and to monitor therapeutic response. Mechanistic pain studies will be foundational to designing these tools and selecting patients for clinical trials of multimodal interventions. This research will guide our understanding of the role for antidepressants and other neuromodulators for people with COP ultimately improving the lives of people with this complex condition.