Is this patient a "responder"? Essential clinical considerations

Other datamethods threads discuss the pitfalls of “responder analysis” from a statistical standpoint. The purpose of this thread is to highlight the type of clinical evidence that’s needed to label a patient a “responder."

If we label a patient a “responder,” we are implying that his therapy has “caused” him to experience a certain effect of interest. In other words, this label requires demonstration of a causal effect of therapy at the level of an individual patient.

The type of evidence needed to call a patient a “responder” will depend on 1) the UNtreated natural history/trajectory of the disease in question; and 2) the expected impact of the therapy being tested.

The solid lines in the five clinical scenarios below show typical UNtreated disease trajectories for various diseases. We’ll imagine what might happen if we were to apply a therapy with intrinsic efficacy/biological activity. The arrow shows the time at which the therapy is applied. The highlighted dotted line after each arrow shows the potential effect of the therapy (with regard to sign/symptom severity/disease activity). The solid line extending past the arrow shows what the patient’s expected clinical trajectory would have looked like if the therapy hadn’t been applied. Examples of conditions that conform to each trajectory are provided, with therapeutic options noted in brackets.

Scenario 1:

IMG_3284

“Waxing and waning” disease course- e.g., asthma (inhalers), chronic pain (analgesics), depression (antidepressants), mild to moderate autoimmune disease (immunosuppressants)

Type of evidence needed to show causality at the level of an individual patient: EITHER an RCT with multiple crossover periods OR an N-of-1 trial is needed. Observing a single period of therapy exposure is INSUFFICIENT to infer therapeutic efficacy for a particular patient.

Rationale; For diseases with a waxing/waning natural history, the causal effect of the intervention, in a specific patient, can only be disentangled from spontaneous improvement/natural fluctuation by observing REPLICATIONS of the effect via therapy dechallenge then rechallenge. Repeated demonstration of a close temporal relationship between therapy dechallenge/rechallenge and change in clinical status isolates the effect of the intervention from the effects of other (often unknown) factors that can contribute to fluctuation in disease course. The waxing/waning of these other factors would not be expected to occur in concert, co-incidentally and repeatedly, with pre-planned periods of therapy dechallenge/rechallenge.

Scenario 2:

IMG_3290

Temporary “slowing” of disease progression, underlying disease is relentlessly progressive- e.g., Alzheimer’s disease (cholinesterase inhibitors)/other neurodegenerative diseases

Type of evidence needed to show causality at the level of an individual patient: Very challenging. Dechallenge/rechallenge is often never performed for fear of triggering more rapid deterioration and because of barriers to accurate assessment of clinical status over time.

Rationale: Therapies for some diseases work by slowing clinical deterioration (at least temporarily), for either all patients or some subset of patients, rather than by improving patients’ clinical state. Identifying “responders” in such scenarios can be very challenging, if not impossible. Causality assessment for individual patients hinges on valid and highly granular mapping of disease trajectory, comparing pre-treatment trajectory with the trajectory during treatment. However, clinical assessments needed to plot trajectory can be confounded by a sometimes “undulating” course of deterioration. For example, scores on cognitive/functional tests for some diseases can vary from day to day or hour to hour for unclear reasons.

Scenario 3:

IMG_3286

Rapid improvement in signs/symptoms very soon after therapy is started, for a disease that has been highly symptomatic for a long period of time- e.g., steroid-dependent autoimmune diseases (asthma/RA/IBD/psoriasis) (biologics)

Type of evidence needed to show causality at the level of an individual patient: The causal effect of therapy is easy to identify clinically for individual patients. Most clinicians would reasonably infer causality even without demonstration of positive dechallenge/rechallenge.

Rationale: Abrupt improvement in the signs/symptoms of a disease that has been highly symptomatic for many years provides clinically compelling evidence that the therapy has caused the patient’s improvement. Therapies that cause such rapid/dramatic improvement are ones that tend to be highly efficacious.

Scenario 4:

IMG_3287

Reduction in disease burden (e.g., tumour burden) soon after start of therapy, for a disease that is known, otherwise, to be relentlessly progressive- e.g., a tumour “melting away” on imaging after starting a new cancer therapy.

Type of evidence needed to show causality at the level of an individual patient: The causal effect of therapy might be easy to identify, clinically, for individual patients (“objective response”), provided techniques for measuring disease burden serially are sufficiently granular/reliable. Serial assessments can be tricky to interpret (e.g., lymph nodes can change slightly in size from scan to scan for reasons unrelated to tumour progression/regression). If scans show a clear trend to improvement in a given patient, dechallenge/rechallenge would not be needed to infer causality.

Rationale: Tumour burden is NOT expected to improve “spontaneously.” Therefore, reduction in tumour burden in a patient after application of a therapy indicates that the effect was caused by the therapy.

Scenario 5:

IMG_3288

Rapid resolution of signs/symptoms soon after the start of a therapy for an acute, highly symptomatic medical condition- such therapies are considered “curative.” e.g., epinephrine for anaphylaxis, primary PCI for STEMI, naloxone for opioid overdose.

Type of evidence needed to show causality at the level of an individual patient: The causal effect of therapy is easy to identify clinically for individual patients. Dechallenge/rechallenge are not needed for the clinician to infer causality.

Rationale: The UNtreated clinical course for many acute conditions is often stereotypical/well-known. Rapid clinical reversal of such conditions is not expected to occur spontaneously. Therefore, abrupt reversal after applying the therapy provides clinically compelling evidence that the therapy caused the reversal/cure.

7 Likes

Related reading to this great post is HDSR’s special issue on “n-of-1” trials.

2 Likes

Really good and appreciated. Perhaps worth a footnote that response to treatment is not evidence of clinical benefit defined as living longer or better. The side effects may offset or be worse than the relief from disease symptoms; the tumor shrinking may be offset by toxicities that shorten life.

2 Likes

Agree completely Karl- I made this point at length in post #18 of the “Dichotomization” thread.

2 Likes

If in some RCT, 0% survive on placebo and 60% survive on treatment, then 60-0 = 60% will have been caused to survive by treatment. Therefore, of the total 60% in the trial surviving on treatment, a total of 60% in the trial were due to treatment. Therefore, the probability that the survival of a patient on treatment in front of us was caused by the treatment is 60/60 = 1.0 as you Erin suggest.

If in some other RCT, 30% survive on placebo and 60% survive on treatment, then 60-30 = 30% will have been caused to survive by treatment. Therefore, of the total 60% in the trial surviving on treatment, a total of 30% in the trial were due to treatment. Therefore, the probability that the survival of a patient on treatment in front of us was caused by the treatment is 30/60 = 0.5.

If in some other RCT, 1% survive on placebo and 60% survive on treatment, then 60-1 = 59% will have been caused to survive by treatment. Therefore, of the total 60% in the trial surviving on treatment, a total of 59% in the trial were due to treatment. Therefore, the probability that the survival of a patient on treatment in front of us was caused by the treatment is 59/60 = 0.98.

It is only when there is 0% survival on placebo that we can be sure that a patient is a responder. If the placebo response is >0%, then all we can do is estimate the probability that the patient was a responder. What do you think?

1 Like

Thanks for your input Huw.

I’m not sure whether the above statement is intended as a general one or specific to the scenario you presented i.e., a universally rapidly fatal disease (?) As a general statement, it would not be true, since, as in Scenarios 3 and 4 above, it would sometimes be possible for us to infer, clinically, that a patient has “responded” to treatment even if an untreated counterpart in the same trial were still alive. Even in Scenario 5, it’s possible that the opioid overdose patient would survive without treatment.

The most salient feature of Scenarios 3-5 that allows us to infer causality for the individual patient is NOT so much the patient’s final OUTCOME in a trial, but the fact that his disease trajectory was altered in a way that would not be expected to occur, clinically, without the application of an intrinsically efficacious therapy.

I know you like math (unlike me :slightly_smiling_face:), but I would be a bit leery about mathematizing these ideas too much. Mathematizing them will tempt mathematically-focused but clinically inexperienced readers to retreat behind their numbers and to miss the main point: Gauging whether it’s possible to assess “response” (in the causal sense of the word) at the level of an individual patient requires knowledge of untreated disease trajectory and assessment of how that trajectory was impacted by treatment. In other words, these assessments require clinical, not simply mathematical, expertise.

1 Like

I agree. My point is that if an outcome never happens without treatment (e.g. survival (as in my example), sudden improvement of long term symptoms (as in your Scenario 3), sudden reduction of tumour burden (as in your Scenario 4) or sudden improvement of severe progressive symptoms (as in your Scenario 5), then if that outcome actually happens on treatment we can infer that the change in outcome was caused by the treatment, as you point out. However, if the above outcomes can occur sometimes or rarely without treatment, then we cannot be sure that the different outcome was due to treatment. In this situation we can only estimate a probability that it was the treatment that caused a better outcome.

If we do an N of 1 trial with multiple crossovers on a single individual, that person may show a ‘good outcome’ on treatment sometimes but a ‘good outcome’ on placebo never. When such a patient gets a good outcome on treatment, we can infer that it was caused by the treatment. However if a patient gets a good outcome on treatment sometimes in the N of 1 trial but also gets a good outcome on placebo occasionally, then we cannot infer for that patient that a good outcome was caused by the treatment; we can only estimate a probability that the outcome was caused by the treatment on that particular occasion.

If a patient obtained a good outcome on treatment more often than on placebo in an N of 1 trial, this would suggest that he or she as an individual would benefit overall from treatment over multiple occasions. We can’t say this for an individual on the basis of a multiple limb RCT; we can only estimate the probability that an individual will benefit overall on repeated occasions from treatment (that person may not actually benefit at all but only others in the trial)…

1 Like

I guess all this is correct in a theoretical sense (?) But the reality is that, after all these years, formal N-of-1 trials are simply not performed in primary care- we’re all just too damn busy! Given the medication list for the average older patient with multiple comorbidities, we’d potentially be running 12 N-of-1 trials per patient if we wanted to convince ourselves that each of his medications was actually (rather than plausibly) helping him (!)

Rather than “proving” incontrovertibly that every medication is making a patient better, most primary care physicians will simply be happy that he’s feeling better. If I do dechallenge/rechallenge for my patients, it’s usually with the goal of establishing whether a physical symptom represents a drug side effect or not (e.g., are the patient’s myalgias due to his statin or not?). Having said this, many primary care physicians (including me) are very vigilant about reassessing medication lists periodically, to see if we can get rid of some of them.

3 Likes

I agree that all this is theoretical. I have never done an N of 1 trial either. However, when patients try to persuade me to give them a treatment from which they are very unlikely to benefit (e.g. testosterone for fatigue with normal endocrinology), I explain that we could do a costly N of 1 trial with multiple cross overs. They invariably decline. However, if patients get better on treatment when the probability of doing so was very low without treatment, I conclude that it was the treatment that probably caused them to get better.

2 Likes

You’re more indulgent than I am :slightly_smiling_face:. I would just tell the patient: “I don’t do that because there’s no evidence for it.” I’d need to see RCT evidence that testosterone supplementation is both effective and safe (over a prolonged period) for treating nonspecific symptoms like fatigue in men with a normal testosterone level before I’d even contemplate doing an N-of-1 trial to see if it helped a specific patient. Of course, I’m not saying that I never prescribe medications “off-label” (all physicians do) but I wouldn’t even consider doing so in the type of situation you’re describing, since I wouldn’t have any evidence about long-term safety.

2 Likes

Well, there was some media publicity some years ago about the decline of testosterone with age in men that caused fatigue (‘male menopause’), which could be improved by administration of testosterone. Some GPs were challenged by some patients to prescribe testosterone even though their testosterone and LH were within the normal range. Because of this, fed-up GPs referred some patients to me. Some of the patients pointed out that the publicity had claimed that there were no RCTs to disprove that testosterone made a difference! I declined to prescribe ‘off label’ explaining that on a priori theoretical grounds, testosterone would not be expected to help if the endocrine results were normal. I explained that such a trial would be turned down probably on cost grounds because of the huge sample size needed to show a difference or no difference. However, I suggested a N of 1 trial could be considered for those prepared to pay or raise funds for it! :slightly_smiling_face:

3 Likes

Yes, I’m very familiar with that saga. It was quite the racket wasn’t it? My approach was: “Just say “no”” ! I was quite familiar with the paucity of evidence at the time, so managed to convince a number of men, but not all. Some patients will never be derailed from their mission to acquire a certain prescription- many of those men probably wound up at one of the “low T” clinics that were popping up left, right, and centre at the time…

2 Likes

Hoping I get to use this logic against Vinay Prasad someday :wink:

2 Likes