Recommendations for cerebrospinal fluid collection for the analysis by ELISA of neurogranin trunc P75, α-synuclein, and total tau in combination with Aß(1-42)/Aß(1-40).

BACKGROUND: The pathophysiology of neurodegeneration is complex. Its diagnosis requires an early identification of sequential changes in several hallmarks in the brains of affected subjects. The presence of brain pathology can be visualized in the cerebrospinal fluid (CSF) by protein profiling. It is clear that the field of Alzheimer's disease (AD) will benefit from an integration of algorithms including CSF concentrations of individual proteins, especially as an aid in clinical decision-making or to improve patient enrolment in clinical trials. The protein profiling approach requires standard operating procedures for collection and storage of CSF which must be easy to integrate into a routine clinical lab environment. Our study provides recommendations for analysis of neurogranin trunc P75, α-synuclein, and tau, in combination with the ratio of ß-amyloid Aß(1-42)/Aß(1-40). METHODS: Protocols for CSF collection were compared with CSF derived from subjects with normal pressure hydrocephalus (n = 19). Variables included recipient type (collection, storage), tube volume, and addition of detergents at the time of collection. CSF biomarker analysis was performed with enzyme-linked immunosorbent assays (ELISAs). Data were analyzed with linear repeated measures and mixed effects models. RESULTS: Adsorption to recipients is lower for neurogranin trunc P75, α-synuclein, and tau (<10%), as compared to Aß(1-42). For neurogranin trunc P75 and total tau, there is still an effect on analyte concentrations as a function of the tube volume. Protocol-related differences for Aß(1-42) can be normalized at the (pre-)analytical level using the ratio Aß(1-42)/Aß(1-40), but not by using the ratio Aß(1-42)/tau. The addition of detergent at the time of collection eliminates differences due to adsorption. CONCLUSIONS: Our study recommends the use of low protein binding tubes for quantification in CSF (without additives) of all relevant CSF biomarkers. Pre-analytical factors have less effect on α-synuclein, neurogranin trunc P75, and total tau, as compared to Aß(1-42). The ratio of Aß(1-42)/Aß(1-40), but not Aß(1-42)/tau, can be used to adjust for pre-analytical differences in analyte concentrations. Our study does not recommend the inclusion of detergents at the time of collection of CSF. The present results provide an experimental basis for new recommendations for parallel analysis of several proteins using one protocol for collection and storage of CSF.

Estimation of diagnostic accuracy of a combination of continuous biomarkers allowing for conditional dependence between the biomarkers and the imperfect reference-test.

Estimating biomarker-index accuracy when only imperfect reference-test information is available is usually performed under the assumption of conditional independence between the biomarker and imperfect reference-test values. We propose to define a latent normally-distributed tolerance-variable underlying the observed dichotomous imperfect reference-test results. Subsequently, we construct a Bayesian latent-class model based on the joint multivariate normal distribution of the latent tolerance and biomarker values, conditional on latent true disease status, which allows accounting for conditional dependence. The accuracy of the continuous biomarker-index is quantified by the AUC of the optimal linear biomarker-combination. Model performance is evaluated by using a simulation study and two sets of data of Alzheimer's disease patients (one from the memory-clinic-based Amsterdam Dementia Cohort and one from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database). Simulation results indicate adequate model performance and bias in estimates of the diagnostic-accuracy measures when the assumption of conditional independence is used when, in fact, it is incorrect. In the considered case studies, conditional dependence between some of the biomarkers and the imperfect reference-test is detected. However, making the conditional independence assumption does not lead to any marked differences in the estimates of diagnostic accuracy.

Optimized Standard Operating Procedures for the Analysis of Cerebrospinal Fluid Aß42 and the Ratios of Aß Isoforms Using Low Protein Binding Tubes.

BACKGROUND: Reduced cerebrospinal fluid (CSF) concentration of amyloid-ß1-42 (Aß1-42) reflects the presence of amyloidopathy in brains of subjects with Alzheimer's disease (AD). OBJECTIVE: To qualify the use of Aß1-42/Aß1-40 for improvement of standard operating procedures (SOP) for measurement of CSF Aß with a focus on CSF collection, storage, and analysis. METHODS: Euroimmun ELISAs for CSF Aß isoforms were used to set up a SOP with respect to recipient properties (low binding, polypropylene), volume of tubes, freeze/thaw cycles, addition of detergents (Triton X-100, Tween-20) in collection or storage tubes or during CSF analysis. Data were analyzed with linear repeated measures and mixed effects models. RESULTS: Optimization of CSF analysis included a pre-wash of recipients (e.g., tubes, 96-well plates) before sample analysis. Using the Aß1-42/Aß1-40 ratio, in contrast to Aß1-42, eliminated effects of tube type, additional freeze/thaw cycles, or effect of CSF volumes for polypropylene storage tubes. 'Low binding' tubes reduced the loss of Aß when aliquoting CSF or in function of additional freeze/thaw cycles. Addition of detergent in CSF collection tubes resulted in an almost complete absence of variation in function of collection procedures, but affected the concentration of Aß isoforms in the immunoassay. CONCLUSION: The ratio of Aß1-42/Aß1-40 is a more robust biomarker than Aß1-42 toward (pre-) analytical interfering factors. Further, 'low binding' recipients and addition of detergent in collection tubes are able to remove effects of SOP-related confounding factors. Integration of the Aß1-42/Aß1-40 ratio and 'low-binding tubes' into guidance criteria may speed up worldwide standardization of CSF biomarker analysis.

Development of a diagnostic test based on multiple continuous biomarkers with an imperfect reference test.

Ignoring the fact that the reference test used to establish the discriminative properties of a combination of diagnostic biomarkers is imperfect can lead to a biased estimate of the diagnostic accuracy of the combination. In this paper, we propose a Bayesian latent-class mixture model to select a combination of biomarkers that maximizes the area under the ROC curve (AUC), while taking into account the imperfect nature of the reference test. In particular, a method for specification of the prior for the mixture component parameters is developed that allows controlling the amount of prior information provided for the AUC. The properties of the model are evaluated by using a simulation study and an application to real data from Alzheimer's disease research. In the simulation study, 100 data sets are simulated for sample sizes ranging from 100 to 600 observations, with a varying correlation between biomarkers. The inclusion of an informative as well as a flat prior for the diagnostic accuracy of the reference test is investigated. In the real-data application, the proposed model was compared with the generally used logistic-regression model that ignores the imperfectness of the reference test. Conditional on the selected sample size and prior distributions, the simulation study results indicate satisfactory performance of the model-based estimates. In particular, the obtained average estimates for all parameters are close to the true values. For the real-data application, AUC estimates for the proposed model are substantially higher than those from the 'traditional' logistic-regression model.

Transferring cut-off values between assays for cerebrospinal fluid Alzheimer's disease biomarkers.

Current technologies quantifying cerebrospinal fluid biomarkers to identify subjects with Alzheimer's disease pathology report different concentrations in function of technology and suffer from between-laboratory variability. Hence, lab- and technology-specific cut-off values are required. It is common practice to establish cut-off values on small datasets and, in the absence of well-characterized samples, to transfer the cut-offs to another assay format using 'side-by-side' testing of samples with both assays. We evaluated the uncertainty in cut-off estimation and the performance of two methods of cut-off transfer by using two clinical datasets and simulated data. The cut-off for the new assay was transferred by applying the commonly-used linear regression approach and a new Bayesian method, which consists of using prior information about the current assay for estimation of the biomarker's distributions for the new assay. Simulations show that cut-offs established with current sample sizes are insufficiently precise and also show the effect of increasing sample sizes on the cut-offs' precision. The Bayesian method results in unbiased and less variable cut-offs with substantially narrower 95% confidence intervals compared to the linear-regression transfer. For the BIODEM datasets, the transferred cut-offs for INNO-BIA Aß1-42 are 167.5 pg/mL (95% credible interval [156.1, 178.0] and 172.8 pg/mL (95% CI [147.6, 179.6]) with Bayesian and linear regression methods, respectively. For the EUROIMMUN assay, the estimated cut-offs are 402.8 pg/mL (95% credible interval [348.0, 473.9]) and 364.4 pg/mL (95% CI [269.7, 426.8]). Sample sizes and statistical methods used to establish and transfer cut-off values have to be carefully considered to guarantee optimal diagnostic performance of biomarkers.

Correcting for the Absence of a Gold Standard Improves Diagnostic Accuracy of Biomarkers in Alzheimer's Disease.

BACKGROUND: Studies investigating the diagnostic accuracy of biomarkers for Alzheimer's disease (AD) are typically performed using the clinical diagnosis or amyloid-ß positron emission tomography as the reference test. However, neither can be considered a gold standard or a perfect reference test for AD. Not accounting for errors in the reference test is known to cause bias in the diagnostic accuracy of biomarkers. OBJECTIVE: To determine the diagnostic accuracy of AD biomarkers while taking the imperfectness of the reference test into account. METHODS: To determine the diagnostic accuracy of AD biomarkers and taking the imperfectness of the reference test into account, we have developed a Bayesian method. This method establishes the biomarkers' true value in predicting the AD-pathology status by combining the reference test and the biomarker data with available information on the reliability of the reference test. The new methodology was applied to two clinical datasets to establish the joint accuracy of three cerebrospinal fluid biomarkers (amyloid-ß 1 - 42, Total tau, and P-tau181p) by including the clinical diagnosis as imperfect reference test into the analysis. RESULTS: The area under the receiver-operating-characteristics curve to discriminate between AD and controls, increases from 0.949 (with 95% credible interval [0.935,0.960]) to 0.990 ([0.985,0.995]) and from 0.870 ([0.817,0.912]) to 0.975 ([0.943,0.990]) for the cohorts, respectively. CONCLUSIONS: Use of the Bayesian methodology enables an improved estimate of the exact diagnostic value of AD biomarkers and overcomes the lack of a gold standard for AD. Using the new method will increase the diagnostic confidence for early stages of AD.

Comparison of two analytical platforms for the clinical qualification of Alzheimer's disease biomarkers in pathologically-confirmed dementia.

Combined analysis of the Alzheimer's disease (AD) biomarkers amyloid-ß(1-42) (Aß(1-42)), total tau (T-tau), and hyperphosphorylated tau (P-tau(181P)) in cerebrospinal fluid (CSF) reduces the uncertainty associated with clinical dementia diagnosis. The present study evaluated the diagnostic accuracy of the CSF biomarker concentrations obtained with a multi-analyte Luminex assay (INNO-BIA AlzBio3) in comparison to single-analyte ELISA tests (INNOTEST). Data from 66 pathologically-confirmed dementia patients (51 AD and 15 non-AD) and 95 controls were included. Cut-off values were determined for each individual biomarker determined using both methods for different diagnostic challenges (dementia-controls; AD-controls; AD-non-AD). Comparing the diagnostic accuracy of individual cut-off values between INNO-BIA and INNOTEST, no relevant differences could be identified. Logistic regression was used in addition to identify the best combination of predictor variables (biomarkers). Discrimination of dementia patients from controls using Aß(1-42) and T-tau yielded a diagnostic accuracy of 0.87 and 0.90 for INNO-BIA and INNOTEST, respectively. Discriminating AD patients from controls, the diagnostic accuracy was 0.90 and 0.93 for INNO-BIA and INNOTEST, respectively. Optimal discrimination of AD and non-AD patients was achieved by combining Aß(1-42) and P-tau(181P) (diagnostic accuracy = 0.86). In conclusion, which AD biomarkers or combination thereof are most informative is dependent on the differential diagnosis, but the clinical value of these markers in each of the differential diagnoses is independent of the method by which concentrations are determined. Since the clinical value of the ELISA (INNOTEST) and Luminex (INNO-BIA) tests is comparable, further research to select the most suitable analytical platform for routine CSF biomarker measurements is needed.

The Alzheimer's Association external quality control program for cerebrospinal fluid biomarkers.

BACKGROUND: The cerebrospinal fluid (CSF) biomarkers amyloid ß (Aß)-42, total-tau (T-tau), and phosphorylated-tau (P-tau) demonstrate good diagnostic accuracy for Alzheimer's disease (AD). However, there are large variations in biomarker measurements between studies, and between and within laboratories. The Alzheimer's Association has initiated a global quality control program to estimate and monitor variability of measurements, quantify batch-to-batch assay variations, and identify sources of variability. In this article, we present the results from the first two rounds of the program. METHODS: The program is open for laboratories using commercially available kits for Aß, T-tau, or P-tau. CSF samples (aliquots of pooled CSF) are sent for analysis several times a year from the Clinical Neurochemistry Laboratory at the Mölndal campus of the University of Gothenburg, Sweden. Each round consists of three quality control samples. RESULTS: Forty laboratories participated. Twenty-six used INNOTEST enzyme-linked immunosorbent assay kits, 14 used Luminex xMAP with the INNO-BIA AlzBio3 kit (both measure Aß-(1-42), P-tau(181P), and T-tau), and 5 used Meso Scale Discovery with the Aß triplex (AßN-42, AßN-40, and AßN-38) or T-tau kits. The total coefficients of variation between the laboratories were 13% to 36%. Five laboratories analyzed the samples six times on different occasions. Within-laboratory precisions differed considerably between biomarkers within individual laboratories. CONCLUSIONS: Measurements of CSF AD biomarkers show large between-laboratory variability, likely caused by factors related to analytical procedures and the analytical kits. Standardization of laboratory procedures and efforts by kit vendors to increase kit performance might lower variability, and will likely increase the usefulness of CSF AD biomarkers.

Validation of ELISA methods for quantification of total tau and phosporylated-tau181 in human cerebrospinal fluid with measurement in specimens from two Alzheimer's disease studies.

Tau measurements in cerebrospinal fluid (CSF) are gaining acceptance as aids to diagnosis of Alzheimer's disease (AD) and differentiation from other dementias. Two ELISA assays, the INNOTEST® hTAU Ag and the INNOTEST® PHOSPHO-TAU(181P) for quantification of t-tau and p-tau181 respectively, have been validated to regulatory standards. Validation parameters were determined by repeated testing of human CSF pools. Specimens from Phase 2 studies of the γ-secretase inhibitor semagacestat and the therapeutic antibody solanezumab at baseline and following 12-14 weeks of treatment were also tested. Estimated intra-assay CV for repeated testing of 3 CSF pools were ≤11.5% and RE varied between -14.1% and +6.4%. Inter-assay CV for t-tau was <5% and RE was within ±8%. For p-tau181, inter-assay CV was <9% and RE was within ±2.5%. Total CV (intra-assay plus inter-assay) were below 10% for both analytes. Up to 20-fold dilutional linearity was demonstrated for both analytes. Stability of t-tau and p-tau181 was demonstrated in CSF during five freeze-thaw cycles at ≤-20 °C and ≤-70 °C and at 18-22 °C for up to 24 h. Neither semagacestat nor solanezumab interfered with either assay. Inter-individual t-tau and p-tau181 concentrations were highly variable but intra-individual variations were small. These assays are suitable for analysis of CSF t-tau and p-tau181 in a single laboratory supporting multi-center AD clinical trials. No effect of treatment with semagacestat or solanezumab was observed in response to three months of drug administration.

Comparison of analytical platforms for cerebrospinal fluid measures of ß-amyloid 1-42, total tau, and p-tau181 for identifying Alzheimer disease amyloid plaque pathology.

BACKGROUND: Cerebrospinal fluid (CSF) biomarkers of Alzheimer disease (AD) are currently being considered for inclusion in revised diagnostic criteria for research and/or clinical purposes to increase the certainty of antemortem diagnosis. OBJECTIVE: To test whether CSF biomarker assays differ in their ability to identify true markers of underlying AD pathology (eg, amyloid plaques and/or neurofibrillary tangles) in living individuals. DESIGN: We compared the performances of the 2 most commonly used platforms, INNOTEST enzyme-linked immunosorbent assay and INNO-BIA AlzBio3, for measurement of CSF ß-amyloid (Aß) and tau proteins to identify the presence of amyloid plaques in a research cohort (n=103). Values obtained for CSF Aß1-42, total tau, and phosphorylated tau 181 (p-tau(181)) using the 2 assay platforms were compared with brain amyloid load as assessed by positron emission tomography using the amyloid imaging agent Pittsburgh compound B. SETTING: The Knight Alzheimer's Disease Research Center at Washington University in St Louis, Missouri. SUBJECTS: Research volunteers who were cognitively normal or had mild to moderate AD dementia. RESULTS: The 2 assay platforms yielded different (approximately 2- to 6-fold) absolute values for the various analytes, but relative values were highly correlated. The CSF Aß1-42 correlated inversely and tau and p-tau(181) correlated positively with the amount of cortical Pittsburgh compound B binding, albeit to differing degrees. Both assays yielded similar patterns of CSF biomarker correlations with amyloid load. The ratios of total tau to Aß1-42 and p-tau(181) to Aß1-42 outperformed any single analyte, including Aß1-42, in discriminating individuals with vs without cortical amyloid. CONCLUSIONS: The INNOTEST and INNO-BIA CSF platforms perform equally well in identifying individuals with underlying amyloid plaque pathology. Differences in absolute values, however, point to the need for assay-specific diagnostic cutoff values.

Qualification of the analytical and clinical performance of CSF biomarker analyses in ADNI.

The close correlation between abnormally low pre-mortem cerebrospinal fluid (CSF) concentrations of amyloid-ß1-42 (Aß(1-42)) and plaque burden measured by amyloid imaging as well as between pathologically increased levels of CSF tau and the extent of neurodegeneration measured by MRI has led to growing interest in using these biomarkers to predict the presence of AD plaque and tangle pathology. A challenge for the widespread use of these CSF biomarkers is the high variability in the assays used to measure these analytes which has been ascribed to multiple pre-analytical and analytical test performance factors. To address this challenge, we conducted a seven-center inter-laboratory standardization study for CSF total tau (t-tau), phospho-tau (p-tau(181)) and Aß(1-42) as part of the Alzheimer's Disease Neuroimaging Initiative (ADNI). Aliquots prepared from five CSF pools assembled from multiple elderly controls (n = 3) and AD patients (n = 2) were the primary test samples analyzed in each of three analytical runs by the participating laboratories using a common batch of research use only immunoassay reagents (INNO-BIA AlzBio3, xMAP technology, from Innogenetics) on the Luminex analytical platform. To account for the combined effects on overall precision of CSF samples (fixed effect), different laboratories and analytical runs (random effects), these data were analyzed by mixed-effects modeling with the following results: within center %CV 95% CI values (mean) of 4.0-6.0% (5.3%) for CSF Aß(1-42); 6.4-6.8% (6.7%) for t-tau and 5.5-18.0% (10.8%) for p-tau(181) and inter-center %CV 95% CI range of 15.9-19.8% (17.9%) for Aß(1-42), 9.6-15.2% (13.1%) for t-tau and 11.3-18.2% (14.6%) for p-tau(181). Long-term experience by the ADNI biomarker core laboratory replicated this degree of within-center precision. Diagnostic threshold CSF concentrations for Aß(1-42) and for the ratio t-tau/Aß(1-42) were determined in an ADNI independent, autopsy-confirmed AD cohort from whom ante-mortem CSF was obtained, and a clinically defined group of cognitively normal controls (NCs) provides statistically significant separation of those who progressed from MCI to AD in the ADNI study. These data suggest that interrogation of ante-mortem CSF in cognitively impaired individuals to determine levels of t-tau, p-tau(181) and Aß(1-42), together with MRI and amyloid imaging biomarkers, could replace autopsy confirmation of AD plaque and tangle pathology as the "gold standard" for the diagnosis of definite AD in the near future.

Diagnosis-independent Alzheimer disease biomarker signature in cognitively normal elderly people.

OBJECTIVE: To identify biomarker patterns typical for Alzheimer disease (AD) in an independent, unsupervised way, without using information on the clinical diagnosis. DESIGN: Mixture modeling approach. SETTING: Alzheimer's Disease Neuroimaging Initiative database. PATIENTS OR OTHER PARTICIPANTS: Cognitively normal persons, patients with AD, and individuals with mild cognitive impairment. MAIN OUTCOME MEASURES: Cerebrospinal fluid-derived beta-amyloid protein 1-42, total tau protein, and phosphorylated tau(181P) protein concentrations were used as biomarkers on a clinically well-characterized data set. The outcome of the qualification analysis was validated on 2 additional data sets, 1 of which was autopsy confirmed. RESULTS: Using the US Alzheimer's Disease Neuroimaging Initiative data set, a cerebrospinal fluid beta-amyloid protein 1-42/phosphorylated tau(181P) biomarker mixture model identified 1 feature linked to AD, while the other matched the "healthy" status. The AD signature was found in 90%, 72%, and 36% of patients in the AD, mild cognitive impairment, and cognitively normal groups, respectively. The cognitively normal group with the AD signature was enriched in apolipoprotein E epsilon4 allele carriers. Results were validated on 2 other data sets. In 1 study consisting of 68 autopsy-confirmed AD cases, 64 of 68 patients (94% sensitivity) were correctly classified with the AD feature. In another data set with patients (n = 57) with mild cognitive impairment followed up for 5 years, the model showed a sensitivity of 100% in patients progressing to AD. CONCLUSIONS: The mixture modeling approach, totally independent of clinical AD diagnosis, correctly classified patients with AD. The unexpected presence of the AD signature in more than one-third of cognitively normal subjects suggests that AD pathology is active and detectable earlier than has heretofore been envisioned.

Evolution of the population structure of Venturia inaequalis, the apple scab fungus, associated with the domestication of its host.

Evaluating the impact of plant domestication on the population structure of the associated pathogens provides an opportunity to increase our understanding of how and why diseases emerge. Here, we investigated the evolution of the population structure of the apple scab fungus Venturia inaequalis in response to the domestication of its host. Inferences were drawn from multilocus microsatellite data obtained from samples collected on (i) the Central Asian Malus sieversii, the main progenitor of apple, (ii) the European crabapple, Malus sylvestris, a secondary progenitor of apple, and (iii) the cultivated apple, Malus x domestica, in orchards from Europe and Central Asia. Using clustering methods, we identified three distinct populations: (i) a large European population on domesticated and wild apples, (ii) a large Central Asian population on domesticated and wild apples in urban and agricultural areas, and (iii) a more geographically restricted population in M. sieversii forests growing in the eastern mountains of Kazakhstan. Unique allele richness and divergence time estimates supported a host-tracking co-evolutionary scenario in which this latter population represents a relict of the ancestral populations from which current populations found in human-managed habitats were derived. Our analyses indicated that the domestication of apple induced a significant change in the genetic differentiation of populations of V. inaequalis in its centre of origin, but had little impact on its population dynamics and mating system. We discuss how the structure of the apple-based agrosystem may have restricted changes in the population structure of the fungus in response to the domestication of its host.

Linked vs unlinked markers: multilocus microsatellite haplotype-sharing as a tool to estimate gene flow and introgression.

We have explored the use of multilocus microsatellite haplotypes to study introgression from cultivated (Malus domestica) into wild apple (Malus sylvestris), and to study gene flow among remnant populations of M. sylvestris. A haplotype consisted of alleles at microsatellite loci along one chromosome. As destruction of haplotypes through recombination occurs much faster than loss of alleles due to genetic drift, the lifespan of a multilocus haplotype is much shorter than that of the underlying alleles. When different populations share the same haplotype, this may indicate recent gene flow between populations. Similarly, haplotypes shared between two species would be a strong signal for introgression. As the expected lifespan of a haplotype depends on the strength of the linkage, the length [in centiMorgans (cM)] of the haplotype shared contains information on the number of generations passed. This application of shared haplotypes is distinct from using haplotype-sharing to detect association between markers and a certain trait. We inferred haplotypes for four to eight microsatellite loci on Linkage Group 10 of apple from genotype data using the program phase, and then identified those haplotypes shared between populations and species. Compared with a Bayesian analysis of unlinked microsatellite loci using the program structure, haplotype-sharing detected a partially different set of putative hybrids. Cultivated haplotypes present in M. sylvestris were short (< 1.5 cM), indicating that introgression had taken place many generations ago, except for two Belgian plants that contained a haplotype of 47.1 cM, indicating recent introgression. In the estimation of gene flow, F(ST) based on unlinked loci indicated small (0.032-0.058) but statistically significant differentiation between some populations only. However, various M. sylvestris haplotypes were shared in nearly all pairwise comparisons of populations, and their length indicated recent gene flow. Hence, all Dutch populations should be considered as one conservation unit. The added value of using sharing of multilocus microsatellite haplotypes as a source of population genetic information is discussed.

Genetic variation in the endangered wild apple (Malus sylvestris (L.) Mill.) in Belgium as revealed by amplified fragment length polymorphism and microsatellite markers.

The genetic variation within and between wild apple samples (Malus sylvestris) and cultivated apple trees was investigated with amplified fragment length polymorphisms (AFLP) and microsatellite markers to develop a conservation genetics programme for the endangered wild apple in Belgium. In total, 76 putative wild apples (originating from Belgium and Germany), six presumed hybrids and 39 cultivars were typed at 12 simple sequence repeats (SSR) and 139 amplified fragment length polymorphism (AFLP) loci. Principal co-ordinate analysis and a model-based clustering method classified the apples into three major gene pools: wild Malus sylvestris genotypes, edible cultivars and ornamental cultivars. All presumed hybrids and two individuals (one Belgian, one German) sampled as M. sylvestris were assigned completely to the edible cultivar gene pool, revealing that cultivated genotypes are present in the wild. However, gene flow between wild and cultivated gene pools is shown to be almost absent, with only three genotypes that showed evidence of admixture between the wild and edible cultivar gene pools. Wild apples sampled in Belgium and Germany constitute gene pools that are clearly differentiated from cultivars and although some geographical pattern of genetic differentiation among wild apple populations exists, most variation is concentrated within samples. Concordant conclusions were obtained from AFLP and SSR markers, which showed highly significant correlations in both among-genotypes and among-samples genetic distances.