New Biomarkers and Biological Therapies in Alzheimer’s Disease

Researchers can analyze biomarkers in both blood plasma and CSF. Traditionally, CSF analysis focuses on three key indicators: levels of Aβ42 (and the Aβ42/Aβ40 ratio), total tau (t-tau), and phosphorylated tau (p-tau). In AD, Aβ42 levels decrease while t-tau and p-tau levels increase—these changes currently form the biological foundation for AD diagnosis [7].

In recent years, researchers have focused particular attention on blood-based biomarkers, as they are less invasive, more cost-effective, and easier to monitor repeatedly than CSF biomarkers. Among these, phosphorylated forms of tau—especially p-tau181 and p-tau217—stand out. Studies have shown that blood levels of p-tau181 correlate with tau and β-amyloid pathology in the brain, making it possible to distinguish AD from other neurodegenerative disorders. Additionally, higher p-tau181 levels are associated with faster cognitive decline and hippocampal atrophy, suggesting its potential prognostic value [8]. p-tau217 demonstrates even higher diagnostic accuracy. Its plasma concentration strongly correlates with tau pathology in both CSF and brain imaging, and its sensitivity and specificity outperform other phosphorylated tau isoforms [9]. Notably, individuals with elevated p-tau181 and p-tau217 levels show a significantly increased risk of developing dementia, particularly before the age of 78, which highlights their potential utility in early diagnosis before clinical symptoms appear [10].

In May 2025, the FDA approved Lumipulse G pTau217/β-Amyloid 1-42, the first commercial blood test for preliminary AD diagnosis. In a clinical study (n = 499), it achieved an AUC of 0.963–0.966, a positive predictive value (PPV) of 91.7%, and a negative predictive value (NPV) of 97.3%, compared to positron emission tomography (PET) or CSF results [11]. Similarly, the Quest AD-Detect™ test (LC-MS/MS) reported an AUC of approximately 0.90, with 100% sensitivity and about 79% specificity relative to amyloid PET imaging (Aβ42/40 cutoff < 0.16) [12–13].

Despite these promising results, clinicians do not yet routinely use blood biomarkers in everyday practice. This is partly due to the absence of standardized pre-analytical protocols and variability in immunoassay methods across laboratories. Moreover, most available data come from highly selected, often homogeneous populations, which limits their generalizability to broader clinical settings.

In addition to established biomarkers, researchers are actively exploring novel indicators to support AD diagnosis. One example is NfL, a marker of axonal damage. Elevated NfL levels in blood correlate with the extent of neurodegeneration. Other emerging candidates include GFAP—a marker of astrocyte activation—as well as YKL-40 and TDP-43, which are linked to neuroinflammation and neuronal degeneration.

For instance, studies have found that patients who later developed dementia had, at baseline, lower Aβ42/40 ratios and higher levels of p-tau181, p-tau217, t-tau, NfL, and GFAP, compared to individuals who remained cognitively stable, including those without AD [10].

However, for most of these biomarkers, clearly defined cutoff values and sufficient data confirming their ability to distinguish AD from other types of dementia are still lacking. Further validation studies are needed in clinically and demographically diverse populations.

Magnetic resonance imaging (MRI), particularly T1-weighted imaging, enables detection of hippocampal and cortical atrophy, changes strongly correlated with the progression of AD. Even in the early stages, such as mild cognitive impairment (MCI), reduced hippocampal and temporal lobe volume can be observed and may serve as a predictor of progression to full-blown dementia. As the disease advances, cortical atrophy extends to additional brain regions, which becomes clearly visible on MRI scans [14–15].

Researchers use advanced MRI analysis techniques, such as volumetry and quantitative volume analysis, not only to monitor the progression of atrophy but also to calculate biometric indicators that support early diagnosis. This approach enables detection of subtle structural brain changes before clinical symptoms become apparent [2].

Another important neuroimaging tool in AD diagnostics is PET, which provides information about cerebral metabolism and the presence of pathological proteins. The most commonly used modality is FDG-PET, which measures glucose uptake. In advanced stages of the disease, reduced glucose metabolism is typically observed, particularly in the temporoparietal and occipital regions [16].

Amyloid PET imaging, which uses labeled ligands (e.g., florbetapir, flutemetamol), allows us to detect β-amyloid plaques. A positive amyloid PET result confirms the presence of amyloid pathology—one of the hallmark features of AD [17].

A newer imaging modality is tau PET, which enables visualization of intracellular neurofibrillary tangles composed of hyperphosphorylated tau protein. By using ligands such as [18F]flortaucipir, researchers can precisely identify affected brain regions, which correlate with the clinical stage of the disease [17–18].

Despite their high diagnostic value, neuroimaging biomarkers are not yet widely used in routine clinical practice due to high costs, limited access to advanced imaging technologies, and the need for expert interpretation. PET imaging also involves exposure to ionizing radiation. Most data come from academic settings, limiting clinical applicability; real-world and cost-effectiveness studies are still needed.

Among known genetic markers, the ε4 allele of the apolipoprotein E (APOE) gene remains the strongest risk factor for late-onset Alzheimer’s disease (LOAD). The APOE gene encodes a protein involved in lipid metabolism and cholesterol transport, and its ε4 variant significantly increases the risk of developing AD. Researchers estimate that 15– 20% of the general population carries at least one copy of the ε4 allele, while its prevalence is much higher among patients with Alzheimer’s disease. Carrying one ε4 allele increases the risk of AD approximately threefold, while homozygosity may raise the risk to 12-fold and is associated with earlier symptom onset [19–21].

Although researchers have not fully clarified the exact mechanism through which APOE ε4 contributes to disease development, studies suggest that it impairs β-amyloid clearance from the brain’s interstitial space, leading to accumulation and plaque formation characteristic of AD. In addition, the ε4 variant enhances inflammatory responses and oxidative stress, thereby accelerating neurodegenerative processes.

In the case of familial Alzheimer’s disease (FAD), which is hereditary, the primary etiological factors include mutations in genes encoding the amyloid precursor protein (APP), presenilin 1 (PSEN1), and presenilin 2 (PSEN2). These mutations lead to excessive production of β-amyloid and increase its neurotoxicity. Although they account for only a small percentage of all AD cases, their presence typically results in very early disease onset, often before the age of 60 [22]. APP mutations directly affect Aβ production, while PSEN1 and PSEN2 mutations impair γ-secretase function, the enzyme responsible for APP proteolysis, thereby favoring production of the highly aggregative Aβ42 form.

Modern genetic research increasingly uses polygenic risk scores (PRS), which combine the effects of multiple genetic variants to estimate an individual’s risk of developing AD. PRS allows scientists to integrate both known and newly identified polymorphisms, offering more accurate risk prediction than any single mutation alone. Studies have demonstrated that higher PRS values correlate with increased likelihood of disease development and with biomarker levels such as Aβ and tau in CSF [22].

Despite their growing diagnostic potential, clinicians have not yet adopted PRS in routine practice. Key challenges include inconsistent methods, limited ethnic diversity in studies, and unclear interpretation at the individual level; further validation and ethical considerations are essential.

To date, most therapeutic strategies evaluated in clinical trials have focused on monoclonal antibodies directed against β-amyloid. Figure 1 illustrates the stages of Aβ aggregation and the binding sites of the antibodies discussed below.

Figure 1.

Mechanisms of Action of Selected Monoclonal Antibodies Against β-Amyloid at Different Stages of Aggregation.

Solanezumab binds Aβ monomers, lecanemab targets protofibrils, gantenerumab binds fibrils and plaques, while aducanumab and donanemab target mature amyloid deposits.

Disclaimer: The diagram provides a conceptual overview; actual binding mechanisms of monoclonal antibodies may involve multiple or variable epitopes not depicted here.

Source: Author’s own elaboration.

Aducanumab was the first antibody to receive accelerated approval from the FDA. This humanized IgG1 monoclonal antibody selectively binds aggregated β-amyloid in the brain. The FDA approved aducanumab in June 2021 based on its ability to target amyloid pathology rather than clinical symptoms alone. Clinical trials showed that aducanumab reduced amyloid plaque burden visible in PET imaging; however, its effects on cognitive function remained inconclusive, sparking widespread scientific debate. Another major concern is its safety profile–aducanumab is associated with a risk of amyloid-related imaging abnormalities (ARIA), including cerebral edema (ARIA-E) and microhemorrhages (ARIA-H), particularly in carriers of the APOE ε4 allele [23–24].

Lecanemab, which targets soluble β-amyloid protofibrils, received FDA approval in January 2023. In February 2025, the European Medicines Agency (EMA) and the Committee for Medicinal Products for Human Use (CHMP) issued a positive opinion, and the European Commission granted formal authorization in April 2025, restricting use to patients carrying no more than one APOE ε4 allele. In the phase III Clarity AD trial (n = 1795), researchers demonstrated that lecanemab moderately slowed cognitive decline— patients showed a –0.45 point difference on the CDR-SB scale after 18 months compared to placebo (95% CI: –0.67 to –0.23; P < 0.001), corresponding to an approximate 27% reduction in progression rate. Treatment also significantly reduced amyloid load as seen on PET imaging. Although ARIA side effects (mainly ARIA-E and ARIA-H) did occur, the frequency was lower than with aducanumab. ARIA-E was observed in 12.6% of patients (symptomatic in 2.8%) [25–29].

Donanemab, one of the most recent anti-amyloid monoclonal antibodies, was evaluated in the large, multicenter TRAILBLAZER-ALZ 2 trial (n = 1736) over 76 weeks. In participants with “intermediate” levels of brain tau, researchers found a 3.25-point reduction in cognitive decline (measured using iADRS) compared to placebo (–6.02 vs. –9.27; 95% CI: 1.88–4.62; P < 0.001), corresponding to a ~35% reduction in decline. In the overall study population, the difference was 2.92 points (95% CI: 1.51–4.33; P < 0.001). Donanemab also showed a significant benefit in functional outcomes, with a 0.67-point improvement on the CDR-SB scale in the low/medium tau group (95% CI: 0.40–0.95; P < 0.001) and a 0.70-point benefit in the full cohort (95% CI: 0.45–0.95; P < 0.001). ARIA-E occurred in 24% of patients receiving donanemab and ARIA-H in 31.3%, compared to 1.9% and 13.0%, respectively, in the placebo group. Symptomatic ARIA-E was reported in 5.8% of treated patients [30–32].

In contrast to aducanumab and lecanemab, which received regulatory approval, other monoclonal antibodies failed to meet clinical expectations. Gantenerumab, designed to bind Aβ aggregates, was tested in two large phase III trials—GRADUATE 1 and GRADUATE 2. Both trials demonstrated significant reductions in amyloid plaque burden (average decrease of 59–63 centiloids; p < 0.001) and favorable changes in fluid biomarkers, including decreased plasma and CSF levels of p-tau217, GFAP, and NfL. However, the clinical impact was statistically insignificant. CDR-SB score changes compared to placebo were –0.14 points in GRADUATE 1 (95% CI: –0.49 to 0.20) and –0.26 in GRADUATE 2 (95% CI: –0.60 to 0.08), both failing to reach statistical significance [33–35].

Solanezumab, which targets monomeric forms of Aβ, showed even weaker effects on amyloid pathology, with PET findings showing no significant changes. In the EXPEDITION 3 trial (n = 2129), the difference in ADAS-Cog14 scores between treatment and placebo groups was just –0.80 points (95% CI: –1.73 to 0.14; P = 0.10), and –0.49 points for MMSE—neither reaching statistical significance. As a result, developers discontinued the drug’s clinical program. Earlier data from EXPEDITION 1 and 2 similarly failed to show cognitive benefits [35–36].

Both solanezumab and gantenerumab were also tested in the DIAN-TU-001 study, the first randomized clinical trial involving patients with dominantly inherited Alzheimer’s disease (DIAD). In this 4-year study (n = 194), neither antibody improved cognitive function or biomarker profiles compared to placebo. While gantenerumab did reduce amyloid burden, this effect did not translate into clinical benefit [33, 37].

These disappointing results emphasize that even substantial reductions in Aβ biomarkers do not necessarily lead to clinical improvement. Factors such as timing of treatment initiation, CNS penetration, mechanistic heterogeneity, and coexisting tau pathology likely influence therapeutic efficacy.

A common limitation of anti-amyloid therapies remains their modest clinical benefit, which, although statistically significant, may be imperceptible to patients and caregivers. Additionally, these treatments are costly, require intravenous administration, and depend on advanced diagnostics such as PET and MRI, limiting their accessibility. Table 3 summarizes the results of the key clinical trials discussed above.

Another direction in the development of Alzheimer’s disease therapies involves immunological strategies targeting tau protein and β-amyloid. In the case of tau, immunotherapy aims to neutralize its pathological forms and inhibit their spread throughout the central nervous system, which could potentially slow neurodegeneration. Several monoclonal antibodies, such as gosuranemab, tilavonemab, and semorinemab, have progressed to clinical trials. These antibodies bind phosphorylated tau and are designed to prevent the formation of neurofibrillary tangles. However, clinical trial results have not demonstrated efficacy in slowing AD progression. Researchers suggest that this may be due to late initiation of treatment, when neurodegeneration is already advanced, or to challenges in appropriately selecting patients, highlighting the need for more precise therapeutic criteria [38–40].

An alternative approach involves active immunization against tau and Aβ using therapeutic vaccines that stimulate the body’s immune response against pathological forms of these proteins. One of the first such candidates was AN1792, a vaccine based on a synthetic Aβ42 peptide. Although early-phase clinical results appeared promising, researchers halted further development after serious adverse events occurred, including encephalitis in approximately 6% of participants [41]. This incident underscored the need for cautious vaccine design and rigorous safety monitoring.

Currently, researchers are developing next-generation vaccines such as ACI-24, UB-311, and ABvac40, which aim to activate the immune system to eliminate amyloid plaques and pathological tau. These candidates are currently undergoing phase II and III trials in larger patient populations. Preliminary data suggest that they may reduce pathological deposits, but their clinical efficacy, especially regarding cognitive improvement, remains uncertain. Furthermore, these vaccines carry a potential risk of inducing neuroinflamma-tory responses, which requires further optimization and thorough evaluation of long-term safety [42–43].

Among novel approaches to AD treatment, several promising strategies are being explored, including ASOs targeting tau mRNA (e.g., MAPTRx/BIIB080) [41], microglial modulation via TREM2 signaling pathways [45–46], and innovative drug delivery systems that bypass the blood–brain barrier, such as intranasal nanocarriers and microneedles [47]. While these methods remain largely experimental, they reflect a growing interest in targeting gene expression, immune processes, and drug delivery challenges in AD therapy. A detailed discussion of these approaches exceeds the scope of this article, but ongoing research will determine their future role in clinical practice.