Fulfilling Expectations

Alzheimer’s disease is not alone. Diabetes, hypertension, and osteoporosis all lack cures, too. Why is it, then, that as a medical community, we are comfortable treating these chronic illnesses associated with aging of the endocrine, cardiovascular, and skeletal systems, but not a progressive, disabling, personally and financially devastating neurological disease like AD? There is no easy answer to that rhetorical question, but the issue has been brought into the spotlight by controversy surrounding two articles about the use of donepezil in the treatment of Alzheimer’s disease. In July 2003, Geldmacher and colleagues¹ published a study—fully funded by Eisai and Pfizer, donepezil’s manufacturing and sales partners—that identified an association between donepezil therapy and the time to nursing home placement. A number of editorialists and correspondents identified methodologic and interpretative issues with that study, and even The New York Times addressed the issue in an article critical of the role of the pharmaceutical industry in such research reports. The criticisms might best be summarized as suggesting that the study was structured and reported in such a way to fulfill the expectations of the manufacturer that donepezil should delay nursing home placement. In June 2004, the results of the AD2000 trial were reported in The Lancet.² This ambitious study—fully funded by the UK’s National Health Service, the primary payor for drug therapy in the UK—set out to identify whether donepezil was cost-effective in the treatment of Alzheimer’s disease. After attempting to include nearly 3000 subjects for participation in a study of 60 or more weeks of donepezil or placebo therapy, the investigators closed enrollment after only 565 patients entered the study. At the end of 60 weeks, only 293—fewer than 10% of the planned enrollment—remained in the study. The investigators reported outcomes out to 3 years of follow-up for nursing home placement, by which time only 20 patients remained enrolled. From the results in this underenrolled and markedly depleted patient cohort, the authors wrote the definitive conclusive statement that “donepezil is not cost effective.” This is a strong statement and a surprising breach of scientific journalism, since even in the nonexistent “perfect” study, the absence of proof cannot be construed as the proof of absence. The nature of that statement further implies that the authors and sponsors wanted it clear that their expectations were fulfilled. A neutral party without expectation of success or failure is unlikely to have phrased the findings so definitively. Given the diametric findings of these two papers, and the apparent fulfillment of opposing expectations, how does a reasonable clinician decide about the benefits of treatment for his or her patient? It is important to remember that efficacy is a statistical construct from the world of clinical trials, and that cost-effectiveness is an economic construct. We, as prescribers, must interpret these constructs on an individual basis for our patients, and determine whether treatment is providing value to the people who take the medicines and those who care for them. One way of conceptualizing value is “improving the odds.” Do acetylcholinesterase inhibitors (AChEIs) improve a patient’s odds of having a better outcome? We know that when untreated, the person with Alzheimer’s disease loses cognitive and functional abilities, though often not in a relentlessly progressive monotonic decline. Numerous, well-conducted double-blind, placebo-controlled studies, as well as the AD 2000 study, show that groups taking AChEIs have better average outcomes on cognitive, functional, and behavioral measures than groups taking placebo. Stated another way, treatment with AChEIs appears to increase the likelihood that the treated person will have better performance at a given point in time. This can be examined in the context of other illnesses like hypertension. In the untreated state, hypertension leads to an increased risk for cerebral infarction, usually in the distant and unpredictable future. Many prescribers will be familiar with a “successfully treated” person with hypertension who has had a stroke. We take for granted that the chronic effects of hypertension will eventually be expressed in many patients, but we also understand from the clinical trials—and much longer clinical experience—that the odds for stroke are reduced by antihypertensive therapy. Why are we dissatisfied with the opportunity to improve the odds in Alzheimer’s therapy? I believe one reason is that we lack a simple predictor of response. It is hard to know when a patient with AD will get the chance to beat the odds of decline, because we don’t have a blood pressure cuff to tell us that the intervention is working on short-term biological intermediaries. Other reasons include the aggressive and overt progression of Alzheimer’s disease symptoms over months, and evolution toward death in about a decade. Therefore, even our successfully treated patient with Alzheimer’s disease is likely to worsen “before our eyes” from one semi-annual visit to the next. Hypertension does its damage silently, inside the blood vessels, over many more years, and we can be lulled into thinking all is well because we don’t see what’s going on. For those troubled by the comparison of treatment of Alzheimer’s disease and the treatment of hypertension, I encourage review of the “point/counterpoint” arguments about the value of treating asymptomatic “benign” hypertension that appeared in the literature through the mid-1960s.^3,4 Another way of assessing the value of Alzheimer’s disease therapy is to find out what is important to patients and their families, and to follow those outcomes. Rockwood and colleagues⁵ did exactly that. In an open-label study, patients’ families were asked to identify their own goals for therapy before beginning donepezil. At regular intervals through the 1-year study, the same family members were queried on whether the patient was improved, stable, or worse on those specific goals. In a robust statement of the value of treatment to this group of families, all goal domains were rated as improved or stable throughout 1 year of therapy. Interestingly, physicians were less satisfied with the outcomes than the patients. Different expectations once again lead to different interpretations of success. Many families identify keeping their loved one with Alzheimer’s disease alive, or out of a nursing home, for an additional year as very-to-extremely important.⁶ Several studies, of varying sponsorship, and varying methods have supported a role for AChEI therapy reducing the risk for, or delaying the time to, nursing home placement among groups of patients with AD.^2,7,8 Will all patients treated with an AChEI avoid nursing home placement for a year? Certainly not, but the odds derived from multiple studies suggest that a significant proportion will. (Exercise: Read the preceding two sentences and substitute “antihypertensive” for “AChEI” and “stroke” for “nursing home placement.” Does it change the fulfillment of your expectation?) Much of the attention paid to AD2000 stems from that study’s failure to find an effect on nursing home placement, but a closer examination of the data, instead of the interpretation, reveals some important methodologic issues. At the end of 1 year 14% of placebo-treated patients were in nursing home care versus 9% of donepezil-treated patients. There was a 15% likelihood that the difference in proportions was attributable to chance (ie, P = 0.15) and—appropriately—no treatment effect was reported. However, when a one-third reduction in the rate of nursing home placement fails to reach statistical significance, the ugly specter of Type II error due to underpowering must be suspected. For example, if only approximately 50% of the targeted enrollment of 3000 subjects had persisted through 1 year of therapy, and the same proportions held, we would be looking at about 67 treated patients versus 105 placebo patients in long-term care. A reduction of that magnitude would have been hard to ignore. Analyses of nursing home placement rates in subsequent years of the AD2000 study lose meaning because of ever-dwindling subject numbers and the effects of an enforced annual treatment washout. One final issue is relative value. That is, is it worth treating Alzheimer’s disease when there are so many other important health issues requiring clinical attention? Every person alive with Alzheimer’s disease will die with Alzheimer’s disease. Many people with Alzheimer’s disease also have hyperlipidemia, and some prescribers feel the need to ration care because of financial constraints. In this situation, it does not make sense to me to defer treating a complicated symptomatic disease associated with clearly reduced quality of life in order to treat an asymptomatic condition that may never cause a problem for that individual. Yet, when prescribers choose a statin in preference to an AChEI in a patient with Alzheimer’s disease, they are making exactly that choice. This point was well articulated by a 2001 Consensus Statement from several leading experts, including a vocal recent critic of the AChEIs, that indicated AChEI drugs were 10-20 times more effective than the common and well-accepted preventive therapies in geriatric medicine, such as antiplatelet or antihypertensive agents.⁹ So, where do we stand with treatment of Alzheimer’s disease today? Our gold standard, double-blind, placebo-controlled trials show that AChEIs provide statistically significant improvements in multiple domains relative to placebo, out to 2 years of treatment. That means they give my patients the best chance to do well in the face of a terrible, dehumanizing disease. AChEIs provide value to many families facing AD, meeting their stated goals, and contribute to delayed (or reduced risk for) nursing home placement. Therefore, I can offer hope to grieving family members and ease their guilt. AChEIs treat what is the most current and most dominant issue for Alzheimer’s disease patients, and I use these medications in preference to systemic preventive treatments when I am forced to recommend a choice. These are the methods I use to fulfill the expectations of those who seek my advice on Alzheimer’s disease. Driven by its own expectations, and complicated by its serious methodologic flaws, I don’t expect the AD2000 study to change my practice.