Why maximum tolerated dose?

Abstract

A long‐established approach to the pharmacological treatment of disease has been to start low and go slow. However, clinicians often prescribe up to maximum tolerated dose (MTD), especially when treating acute and more severe disease, without evidence to show that MTD is more likely to improve outcomes. Cardiovascular guidelines for some indications advocate MTD even in prevention, for example hypercholesterolaemia, without compelling evidence of better outcomes. This review explores the origins and potential problems of prescribing medications at MTD. Oral effective dose 50 (ED50) may be a useful guide for balancing efficacy and safety.

1. INTRODUCTION

The common practice of prescribing the maximum tolerated dose (MTD) of drugs such as antidepressants, antipsychotics, stimulants, statins,1, 2 vasodilators and β‐blockers3 is not evidence based. The origins of MTD can be traced to animal toxicity testing in drug development4 after a pharmacologically active dose has been established.5 Different definitions of MTD have emerged4 since Sontag's original proposal: “the highest dose of the test agent given during the chronic study that can be predicted not to alter the animals’ normal longevity from effects other than carcinogenicity.”6 Early phase human clinical trials attempted to establish the highest dose of a drug that does not cause overt toxicity or unacceptable adverse effects within a specified time period.

Clinicians often base dose on disease severity and MTD may be presumed necessary or worth trialling in serious or worsening conditions such as malignant and cardiovascular disease, vasculitis, epilepsy and serious mental illness. However, there is no evidence for better clinical outcomes with MTD and harms may overtake benefit as dose is increased. Continuation of a dose increase is justified only if there is unequivocal evidence of improved efficacy, along with acceptable adverse effects, especially in long‐term treatment of chronic disease.

Benefit may be difficult to weigh against the risk of harms with dose increases up to MTD. Effective dose 50 (ED50) defines the centre of the dose response curve, the mean population dose required to achieve half the maximum drug effect, 50% of maximum effective dose.7 ED50 may be based on measurement of biochemical or clinical endpoints and may be a useful guide to clinicians. ED50 may not be well defined in relation to clinical outcomes as, in much drug development, one dose becomes favoured in Phase I and later human trials. Investigation of the effects and outcomes on at least 3 different doses is required to estimate ED50.

With statins, coronary event reduction plateaus just above ED50 whilst a range of adverse drug reactions (ADRs) continue to increase.8 Lipid guidelines in North America have noted since 2013 that cholesterol targets have no evidence base,1 yet along with European guidelines2 promote MTD as a therapeutic target despite known reduced statin adherence related to adverse effects,9 which have been shown to be dose‐related.8 The decline in the mortality benefit on higher doses of statins8 suggests that some ADRs may contribute to mortality, for example from liver dysfunction, rhabdomyolysis, renal impairment in patients with pre‐existent renal disease10 and cerebral haemorrhage, especially in patients with previous stroke,11 which is of particular concern in the elderly.

Antihypertensive drugs are often used at higher doses than were employed in RCTs (Table 1), and aimed at epidemiological targets, which fail to factor in adverse effects of drugs. Heart failure guidelines recommend MTD of angiotensin‐converting enzyme inhibitors, angiotensin receptor blockers and β‐blockers,3 for example up to 7‐fold the ED50 of metoprolol and 16‐fold the ED50 of candesartan, despite the inevitable increase in ADRs. Cross‐sectional observational data may appear to suggest greater benefit with higher dosing but are potentially confounded by the greater tolerance of higher doses in patients with less severe disease. In contrast, no reduction in total mortality, which provides a useful summary of efficacy and safety, is evident when dose allocation (to higher or lower doses) of statins8 and angiotensin‐converting enzyme inhibitors, angiotensin receptor blockers or β‐blockers was randomised.20, 21

Table 1.

Approved, maximum and ED50 doses of common central nervous system and cardiovascular system drugs

Daily dose
Drug	Maximum reported (mg)	Approved	Mean population
		Oral ED50 dose (estimated)
		(mg)	(mg)
		(as ratio of ED50)
Selective serotonin reuptake inhibitors 12
Citalopram	80	10–40 (3–14)	3
Sertraline	200	50–100 (5–11)	9
Fluoxetine	80	10–60 (3–20)	3
Paroxetine	60	10–40 (2–8)	5
Venlafaxine	400	37.5–225 (6–38)	6
Antipsychotics 13
Chlorpromazine	2000	25–100 (0.2–0.7)	150
Haloperidol	20	0.5–2.5 (0.5–2.5)	1
Quetiapine	800	25–200 (0.2–1.3)	150
Risperidone	16	0.5–2 (0.2–1.0)	2
Olanzapine	60	2.5–20 (0.3–2.2)	9
Stimulants
Methylphenidate 14	100	5–60 (0.5–6)	10
Dexamphetamine 15	40	5–20 (0.5–2)	10
Beta‐blockers 16
Propranolol	320	10–160 (0.2–4)	40
Metoprolol	400	50–100 (0.8–2.2)	60
Diuretics
Hydrochlorothiazide 14	100	12.5–50 (1.2–5)	10
Frusemide 12	500	20–40 (2–4)	10
Angiotensin‐converting enzyme inhibitors 17
Captopril	300	12.5–50 (0.6–2.5)	20
Perindopril	32	2–8 (0.5–2.0)	4
Statins 8
Simvastatin	160	10–80 (0.6–5)	15
Atorvastatin	80	10–80 (5–40)	2
Rosuvastatin	40	5–40 (5–40)	1
Warfarin 18	20	2–8 a (0.3–1.3)	6 b
Thyroxine 19	0.2	0.05–0.1 (0.7–1.3)	0.075 c

^a25^th–75^th centile dose.

^bmedian population dose.

^cmean population dose.

2. ADRS

Drugs usually target one benefit but may cause or contribute to a variety of adverse effects. Efficacy plateaus above ED50, whereas the incidence and severity of ADRs generally increase with dose,22 [Ferner, Aronson, this issue BJCP], exponentially with anticoagulants.23 Some ADRs are related to known drug actions, for example related to hypotension in the treatment of blood pressure or hypoglycaemia in the treatment of diabetes. However, many ADRs are less well understood. Toxicity symptoms may be delayed22 or subtle, for example, liver and renal toxicity. Clinicians should always be on high alert for ADRs, as prescribed drugs contribute to or cause 7–16% of deaths in hospitalised adults.24, 25, 26, 27

3. CLINICAL DOSING

Optimal dosing aims to maximize benefit and minimize harm and may vary between patients in acute, maintenance and preventive therapy. The temptation to increase to MTD arises in acute, serious and severe illness. Symptoms, vital signs and other surrogate indicators can provide a guide to titrating dose in both clinical trials and practice. Difficulties arise when dosing is based purely on symptoms such as pain or reported distress, commonly in psychiatry. Without objective indicators against which to titrate, doses vary widely based on a clinician's impression of dose–response. For example, whilst the recommended daily dose of desvenlafaxine is 50 mg, clinicians have prescribed up to 400 mg, many‐fold ED50 (Table 1), despite the evidence of increased ADRs28 and no reliable evidence of additional benefits. The failure to demonstrate greater clinical effectiveness with higher doses of first‐generation antipsychotics and the high rates of serious adverse effects drove their dose reduction (Table 1) and the development of second‐generation agents. However, second generation antipsychotics have also been prescribed above the recommended dose, for example up to 60 mg daily of olanzapine (Table 1), despite well documented cardiovascular29, 30 and other adverse effects.30

Clinical trials show that antidepressants are no better than placebo for mild depression and only marginally better than placebo for more serious depression27 but are often prescribed at many‐fold their ED50 (Table 1). There is growing recognition that between 20 and 30% of more serious mental illness does not respond appreciably to pharmacotherapy.31, 32 This has important implications for dosing. With only clinical indicators as a guide, clinicians may continue to increase the dose in unresponsive patients in the hope of a response that may not eventuate. There is little evidence of greater efficacy with higher doses of antidepressants33 or antipsychotics,34 whereas ADRs generally continue to increase with dose.22 Adverse effects reduce adherence.35 Patients should probably not be left on psychotropics, particularly at higher dose, without a trial of lower doses or deprescribing.36 Close consideration should be given to the use of antipsychotics in the elderly and in dementia, particularly their off‐label use for sedation and behavioural disturbance.37 Published data show that compared to placebo, harms exceed benefit38 and mortality is higher39 on antipsychotics.

An established approach to dosing, particularly for chronic diseases such as hypertension, is to start low and go slow,40 especially in the elderly. This approach facilitates dose titration to optimize effectiveness and tolerability. However, less cautious approaches may be adopted because of patient, family and community expectations of rapid results, which may drive therapeutic impatience and commencement of higher drug doses.

Safety concerns in the post‐marketing period with widely used medications have led to proactive approaches to drug safety41 such as Risk Evaluation and Mitigation Strategies42 and risk management plans (RMPs)43 for newer drugs at the time of entry to the market (Food and Drug Administration, European Medicines Agency). However, many postmarketing safety issues cannot be identified or anticipated in RMPs, with almost half of the variations in product information concerning safety aspects not envisaged by the RMP.44 These approaches may paradoxically promote a false sense of safety and lead physicians to think less critically about tempering dose.

4. DOSING AND OUTCOMES

There is a paucity of reliable data on dose‐related outcomes, even for commonly prescribed drugs. Safety and the relationship of tolerability to compliance in chronic disease has driven research in some areas to establish essential or sufficient dose, between pharmacologically active dose and no observable adverse effect level.5 Much researched examples include low‐dose aspirin45 for the prevention of stroke and coronary events and low dose methotrexate for rheumatoid arthritis,46 [Lucas, Dimmitt, Martin, this themed issue BJCP]. Metformin and paracetamol (both with ED50s around 2 g) are never recommended above their ED50. Colchicine, used in acute gout for many years with limited evidence on which to base dosing47 is a good historical example of MTD dosing. Earlier dosing regimens consisted of an immediate dose (1 mg), the usual dose threshold, with continued dosing (0.5 mg) at 1–2‐hourly intervals until the development of toxicity (diarrhoea, nausea), or symptom relief. However, the first controlled study in acute gout48 demonstrated toxicity in all patients given colchicine (mean dose 6.7 mg) with diarrhoea after a median of 24 hours, usually before the benefit of pain relief emerged. A subsequent dose comparison study demonstrated no greater benefit but only increased harms with high‐dose (4.8 mg) compared with low‐dose (1.8 mg) colchicine.49

Many patients have been prescribed up to 500 mg frusemide daily (Table 1) although safety and clinical benefits above 80 mg daily are dubious. The mean population oral ED50 is <10 mg daily.50 Higher doses do not appear to confer greater benefit even in heart failure51 and particularly with the advent of other effective coadministered antifailure medications, most patients are controlled with smaller doses of frusemide, for example 20–40 mg daily. Thiazide diuretics can cause hypokalaemia, hyponatraemia, renal impairment and increase the risk of gout and diabetes, all at least partly dose related. Doses of hydrochlorothiazide have consequently fallen from up to 100 mg daily down to 6.25–12.5 mg,52 which is around the ED50 (Table 1). Beta‐blockers confer most benefits in left ventricular failure at doses around or below ED50,21 for example 25–50 mg of metoprolol. Higher doses are less well tolerated and slower heart rate can increase systolic blood pressure, potentially negating benefits.21

The imperative to prescribe doses near or up to MTD in symptomatic and ill patients is understandable but is not supported in most instances by reliable evidence and particularly in the elderly, is unwise. More research on dose‐related outcomes is required to establish guidelines that clinicians can adopt more confidently.

5. CONCLUSION

Higher drug doses may be necessary in acute or severe illness but, when patients are stable, lower doses often suffice to ensure effective maintenance treatment, tolerability, minimisation of adverse effects and long‐term adherence. MTD as established in drug development prior to human clinical trials appears likely to come with unacceptable risks and little gain if used as a clinical dosing strategy. MTD is not justified in prevention where any adverse effects may be unacceptable. Without reliable evidence of the optimal population dose with respect to clinical outcomes and survival, clinicians may be left with limited guidance for managing patients. Mean population ED50 may be a reasonable starting dose, with due allowance for body size. The dose can be increased above ED50 if greater efficacy can be demonstrated and lowered should ADRs develop. In most illness, there is merit in maintaining patients on the lowest effective dose, with close clinical monitoring for efficacy, tolerability and adverse events.

COMPETING INTERESTS

There are no competing interests to declare.

Stampfer HG, Gabb GM, Dimmitt SB. Why maximum tolerated dose? Br J Clin Pharmacol. 2019; 85: 2213–2217. 10.1111/bcp.14032