Introduction
Alzheimer's disease (AD) is the commonest form of dementia, estimated to affect 37 million people worldwideReference Caraci, Copani, Nicoletti and Drago1 and over 400,000 in the UK.2 It is associated with the build up of plaques in the brain leading to neuronal cell death.Reference Dawbarn and Allen3 The most common early symptom of AD is memory loss, but it leads to progressive cognitive impairment, behavioural and psychological change and increasing dependency.4 The cause of AD is unknown, although the largest risk factor for its development is age, with over 15% of people aged over 80 in the UK having AD.2
There are currently no drugs available to halt progression of AD.Reference Lleo, Greenberg and Growdon5 Cholinesterase inhibitors, the current drugs of choice, slow the development of symptoms. However, they are not effective for everyone and have decreased efficacy over time.6 Alternatively, disease-modifying interventions such as immunotherapy aim to halt or even reverse symptoms like memory loss by removing β-amyloid (Aβ) pathology. The ‘amyloid hypothesis’ of AD suggests an imbalance in the production and clearance of Aβ1–42, leading to accumulation and aggregation of amyloid plaques in the brain.Reference Hardy and Selkoe7 Evidence indicates that Aβ is neurotoxic and its accumulation initiates a cascade of events leading to synaptic failure and neuronal death.Reference Selkoe8
Interest in immunotherapy has mainly focused on two areas: firstly, active immunization whereby Aβ is used as an antigen and the vaccinated patient has to produce antibodies against it; and secondly, passive immunization, where pre-prepared anti-Aβ antibodies are injected.
This review is informed by a literature search of Medline and Scopus that was conducted in March 2010. Search terms included: amyloid, Alzheimer's disease, immunotherapy. Fuller details of the search methods are available from the author.
Active immunization in an animal model
In 1999, a paper was produced by Schenk et al.Reference Schenk, Barbour, Dunn, Gordon, Grajeda, Guido, Hu, Huang, Johnson-Wood, Khan, Kholodenko, Lee, Liao, Lieberburg, Motter, Mutter, Soriano, Shopp, Vasquez, Vandevert, Walker, Wogulis, Yednock, Games and Seubert9 documenting the findings in transgenic mice models of AD immunized with Aβ. These transgenic PDAPP mice over-express Aβ42, the predominant form of Aβ found in plaques. Two subsets of mice were examined: 6-week-old mice that did not yet show evidence of AD-like pathology, and 11-month-old mice with evidence of AD-like pathology. The mice were split into two groups: a control group immunized with phosphate-buffered saline or untreated, and a group immunized with Aβ42. The findings were that a statistically significant proportion of younger mice had a reduced number of plaques compared with the control group. In the treated group, seven of the nine mice had no Aβ plaques remaining. The older control and immunized mice were killed at 12, 15 and 18 months and their brains were analysed for evidence of plaques. Immunized mice had a reduced Aβ burden of 0.01% at 18 months, compared with 4.87% in the control group. A key finding from this study was that Aβ42 immunization resulted in generation of anti-Aβ antibodies. Eighteen months post-completion it was reported that the expected cognitive decline in these mice was halted.Reference Janus, Pearson, McLaurin, Mathews, Jiang, Schmidt, Chishti, Horne, Heslin, French, Mount, Nixon, Mercken, Bergeron, Fraser, St George-Hyslop and Westaway10
This was the first clinically relevant treatment demonstrating a significant reduction of plaques that would otherwise have occurred in the PDAPP mouse. Although results were promising, they were from mice, not human patients. Humans have more complex brains than mice and so there is possibly more potential for adverse reactions. However, the positive findings were considered sufficient to justify human trials.
Active immunization in humans using AN1792
Phase II clinical trials began to investigate the potential vaccine Aβ42 (AN1792) in humans. Following an initial dose ascending trial in 80 patients,Reference Bayer, Bullock, Jones, Wilkinson, Paterson, Jenkins, Millais and Donoghue11 a larger study was begun to investigate the safety, tolerability and efficacy of the vaccine in a randomized, multiple-dose, double-blind trial.Reference Gilman, Koller, Black, Jenkins, Griffith, Fox, Eisner, Kirby, Rovira, Forette and Orgogozo12 The latter study included 372 patients aged 50–85 years, all with AD and from 28 centres in Europe and the USA. There were 300 patients in the active treatment group and 72 in the control group receiving a placebo. The trial was scheduled for 15 months and injections were administered intramuscularly on the first day of the trial and at 1 and 3 months. However, the trial was halted early due to development of encephalitis in 6% of patients in the treatment group.Reference Orgogozo, Gilman, Dartigues, Laurent, Puel, Kirby, Jouanny, Dubois, Eisner, Flitman, Michel, Boada, Frank and Hock13 Despite not receiving all of the planned vaccinations, 19.7% of the treatment group were antibody responders. However, five out of the six tests used to assess cognition showed no significant difference in scores between the antibody responder group and the placebo group. The only significant difference was in the Neuropsychological Test Battery (NTB) score, a composite cognitive score, where antibody responders performed better than the control group. Antibody responders also had increased losses in brain volume measured by serial MRI scan, possibly due to Aβ removal and associated cerebral fluid shifts.Reference Fox, Black, Gilman, Rossor, Griffith, Jenkins and Koller14 Treatment-related adverse events were more common on active treatment than in controls, but were generally of mild to moderate intensity. However, all 18 patients who reported meningoencephalitis were in the treatment group and this was probably associated with a T-cell response.Reference Weksler15
Follow-up of AN1792
In 2008, long-term data were published on the clinical efficacy, survival rates and cause of death in patients immunized with AN1792.Reference Holmes, Boche, Wilkinson, Yadegarfar, Hopkins, Bayer, Jones, Bullock, Love, Neal, Zotova and Nicoll16 The original trial had lasted 84 weeks and involved 64 patients who received AN1792, and 16 controls, all with mild to moderate AD according to standard clinical criteria.Reference Bayer, Bullock, Jones, Wilkinson, Paterson, Jenkins, Millais and Donoghue11 A year after initial study completion, ethical approval was obtained to contact all surviving patients for follow-up at yearly intervals until completion of the study or death. Twenty-four patients and/or carers refused consent for clinical follow-up, a limitation of the study. Importantly, consent also was sought for post-mortem neuropathological examination in the event of death and this led to brain donation from eight patients who had been immunized. As no unimmunized placebo patient had a post-mortem, there was a need for an alternative unimmunized control group for histological comparisons of AD load. These control cases, matched to the immunized group in terms of age at death, were obtained from the neuropathology archives of Southampton General Hospital. Various areas of the brains were examined histologically for the presence of Aβ, with analysis in each case carried out in the same laboratory and by the same neuropathologist, allowing for consistency within a single controlled environment. The authors recognized a limitation in the technique of calculating the percentage of Aβ load, as it has ‘considerable variation in AD’Reference Holmes, Boche, Wilkinson, Yadegarfar, Hopkins, Bayer, Jones, Bullock, Love, Neal, Zotova and Nicoll16 and as baseline values were not recorded, assessing change in individual patients was not possible.
The crucial finding of this study was that mean Aβ load was significantly lower in immunized patients than in unimmunized controls. The degree of plaque removal also varied with the mean antibody response during the study. However, the limited number of post-mortem cases means the findings should be interpreted cautiously. There was little evidence to suggest any major effect on cognitive function. There was no difference between the treatment and control groups in timing of progression to severe AD and the 26 surviving patients displayed no significant difference in cognitive and functional scores when compared with the placebo group, suggesting plaque removal itself may not be enough to limit cognitive decline. The authors concluded that there was ‘no evidence of immunization having any influence of long-term survival or clinical outcomes’.Reference Holmes, Boche, Wilkinson, Yadegarfar, Hopkins, Bayer, Jones, Bullock, Love, Neal, Zotova and Nicoll16 These results suggest removal of plaques by active immunization may not be enough to halt neurodegeneration in AD.
Passive immunization in a transgenic mouse model
Passive immunization with pre-prepared antibodies is an attractive alternative to active vaccination. It can incorporate successful elements of active vaccination without its adverse side-effectsReference Foster, Verdile, Bates and Martins17 and has been given support by findings in animal models that IVIg helps clear Aβ from the brain.Reference Istrin, Bosis and Solomon18
One of the first studies to show that passive immunization cleared Aβ plaques from the brain was published by Bard et al. in 2000.Reference Bard, Cannon, Barbour, Burke, Games, Grajeda, Guido, Hu, Huang, Johnson-Wood, Khan, Kholodenko, Lee, Lieberburg, Motter, Nguyen, Soriano, Vasquez, Weiss, Welch, Seubert, Schenk and Yednock19 Like the Schenk study,Reference Schenk, Barbour, Dunn, Gordon, Grajeda, Guido, Hu, Huang, Johnson-Wood, Khan, Kholodenko, Lee, Liao, Lieberburg, Motter, Mutter, Soriano, Shopp, Vasquez, Vandevert, Walker, Wogulis, Yednock, Games and Seubert9 this trial used APP transgenic mice which over-express Aβ42, and show pathological features of AD. Eight mice were give intraperitoneal injections of either monoclonal (10D5 or 21F12) anti-Aβ antibodies, or a polyclonal immunoglobulin (IVIg), weekly for 6 months. Quantitative image analysis was used to determine Aβ burden and enzyme-linked immunoassay (ELISA) was used to determine Aβ42 levels in the cortex. The results were promising, with a 93% reduction in plaque burden in the mice immunized with Ig and 81% reduction with monoclonal antibody 10D5. Additionally there were significantly reduced levels of cortical Aβ42 by ELISA measurements in both Ig and 10D5 groups. Mice immunized with 21F12 monoclonal antibody had high plaque burden and no substantial reduction in ELISA score.
In a separate study, additional antibodies against Aβ 3D6 and 16C11 were tested. The study aim and design was the same as before, except the mice were older, aged between 11.5 and 12 months. After 6 months, 3D6 was effective in reducing plaque burden by 86% when compared with the control group and 16C11 was ineffective. However, the key finding was that no T-cell response was demonstrated, unlike the occasional findings in active immunization. At the end of the study, the mouse brains were examined microscopically. Antibodies against Aβ had entered the brain, but after 35 days there was no obvious removal of large plaques. However, there did appear to be a reduction in small aggregates of Aβ.
Passive immunization in humans
The first trial by Dodel et al. investigating passive immunization in humans was published in 2004.Reference Dodel, Du, Depboylu, Hampel, Frölich, Haag, Hemmeter, Paulsen, Teipel, Brettschneider, Spottke, Nölker, Möller, Wei, Farlow, Sommer and Oertel20 This study used antibodies against Aβ found in human IVIg preparations, with the aim of improving cognitive function in five patients with ‘clinically probable or possible AD’. Immunoglobulin was given on three consecutive days every 4 weeks for 6 months, at a total dose of 0.4 g/kg of body weight. The safety of the immunoglobulin was assessed at every visit and efficacy by measuring Aβ levels in CSF and serum at baseline and 6 months, in addition to cognitive testing. The results showed the concentration of Aβ in the CSF decreased for all patients over the duration of the study. In addition, a slight improvement on cognitive testing was recorded at 6 months in four out of the five patients, though there was considerable variance in the cognitive scores. No patients experienced adverse effects related to the medication, offering promise for the immunoglobulin to be studied in a larger cohort of patients. Furthermore, decline on MMSE was slower in patients treated with the immunoglobulin (3.7 points) compared with patients treated with cholinesterase inhibitors over the same time period (4–6 points).
A study by Relkin et al. investigating the safety of repeated IVIg infusions and the effect dose has on levels of Aβ in the plasma and CSF over an 18-month period was published in 2009.Reference Relkin, Szabo, Adamiak, Burgut, Monthe, Lent, Younkin, Younkin, Schiff and Weksler21 In this study, the patients were older and the trial period was longer than the initial study by Dodel et al. Eight patients with a mean age of 74.3 years, all with probable mild AD, were included. A variety of markers including IGA levels, renal function and metabolic profile were assessed at baseline and throughout to assess safety. Efficacy was tested using the MMSE at baseline and after each infusion to monitor progress, in addition to recording blood and CSF measures at baseline and at 3-monthly intervals thereafter. Patients were randomly assigned to one of four different IVIg doses (but with no untreated control group) for 6 months and then stopped treatment for 3 months, before recommencing for a further 9 months.
After a test dose of IVIg, all eight patients had detectable anti-Aβ antibodies and the increases in titres were significant, with a positive correlation between the amount of IVIg infused and the titre of serum anti-Aβ antibody. After 6 months of treatment with IVIg, Aβ levels in the CSF had significantly decreased, but when treatment was stopped during the 3-month wash-out period, levels of Aβ returned to the original levels, but decreased again when treatment was recommenced. Six of the eight patients remained at or above the original MMSE score at the end of the study. However, the open label design of the study may have introduced bias as both patients and researchers knew that active drug was being given.
Another phase II trial has tested the efficacy and safety of bapineuzumab, a monoclonal antibody thought to bind to Aβ in the brain and facilitate its removal.Reference Salloway, Sperling, Gilman, Fox, Blennow, Raskind, Sabbagh, Honig, Doody, van Dyck, Mulnard, Barakos, Gregg, Liu, Lieberburg, Schenk, Black and Grundman22 The study was conducted at thirty different sites across the US, with strict inclusion and exclusion criteria. The patients were randomly assigned to receive one of four doses of bapineuzumab (n = 124) or placebo (n = 110), administered intravenously over 1 hour, every 13 weeks for 18 months. Only 65% of the treatment group completed the study. At completion, there were statistically significant differences in three out of the four treatment groups in the cognition tests. Further analysis of results was based on the APOE4 status of the participants. The 146 carriers of APOE4 in the study showed no difference at the end-points for any of the cognitive tests when compared with the control group, whereas in the 79 non-carriers, differences were noted on the ADAS-cog and MMSE. However, these post-hoc results should be interpreted cautiously, due to a higher proportion of non-carriers completing the study, as well as the more advanced Aβ pathology in APOE4 carriers potentially affecting results.
The drug was generally well tolerated. However, vasogenic oedema occurred in 9.7% of the active treatment group who received doses of bapineuzumab >1.0 mg/kg. Patients who were in the dosing group of 0.5 mg/kg did not experience any vasogenic oedema. There have also been additional studies reporting microhaemorrhages following passive vaccination.Reference Hawkes and McLaurin23 There has been a recent report of a reduction in Aβ load on 11C-PiB PET scans in AD patients after bapineuzumab treatment, compared with both baseline and placebo.Reference Rinne, Brooks, Rossor, Fox, Bullock, Klunk, Mathis, Blennow, Barakos, Okello, Rodriguez Martinez de Liano, Liu, Koller, Gregg, Schenk, Black and Grundman24 Phase 3 trials are in progress.
Conclusion
It is just over 10 years since Schenk et al. showed that the immune system could be harnessed against AD. Unfortunately, rapid translation of the findings in the murine model to a human trial resulted in encephalitis in 6% of actively immunized patients.Reference Town25 This highlights the issue that new treatments for AD should not be tested on humans until they have been tested on non-murine species such as primates. Although active and passive mechanisms both aim to clear Aβ levels in the brain and appeared successful in murine models, passive immunization seems a more attractive option. IVIg has been used for several decades in other treatments, with good clinical tolerability and an established safety record.Reference Foster, Verdile, Bates and Martins17 Additionally, compared with active vaccination there is greater control over the anti-Aβ titres and the ability to stop treatment rapidly if adverse events occur.Reference Hawkes and McLaurin23
A recurring theme in studies to date has been small study size and the use of different methods to assess cognitive ability, making comparison difficult. The necessarily narrow inclusion criteria for the trials have also limited their generalizability to the ‘real world’. Crucially, some patients’ clearance of Aβ plaques did not necessarily result in improvements in cognition, perhaps suggesting that removal of Aβ in itself was not sufficient to improve cognitive function. In future studies it may be necessary to target the point in the amyloid cascade prior to plaque formation,Reference Hawkes and McLaurin23 perhaps treating patients whilst still asymptomatic. Ensuring adequate participant numbers and consideration of their APOE4 status is important.
The cost of immunotherapy may be thousands of pounds per dose and if proven to be of value, routine usage is unlikely to be feasible.Reference Brody and Holtzman26 This is still a developing field and a number of phase II–III clinical trials are currently underway.Reference Hawkes and McLaurin23 Their results are awaited with much anticipation.
Conflicts of interest
None to declare.
This article is based on an essay submitted for the Amulree Prize of the British Geriatric Society in 2010.
