Impact of mammographic screening on the detection of good and poor prognosis breast cancers

We sought to compare the molecular signature of node negative cancers from two cohorts 15 years apart, to determine if there is molecular evidence of increase in low and ultralow risk cancers over time. We studied the impact of age, time period of diagnosis, and mammographic screening on biology of tumors where The Netherlands Cancer Institute 70-gene prognosis signature was generated as part of 2 validation series, one retrospective (1984-1992), Cohort 1, and one prospective (2004-2006), Cohort 2. A total of 866 patients were analyzed. Regardless of time period of diagnosis, the proportion of T1, grade 1, hormone receptor positive (HR) tumors, and good prognosis by 70-gene signature significantly increases as age increases (P < 0.01). In women aged 49-60, the time period of diagnosis significantly affects the proportion of cancers that were NKI 70-gene low risk: 40.6% (67/165) compared with 58% (119/205) for Cohorts 1 and 2, respectively. This is in contrast to the absence of a significant change for women under age 40, where 25% (17/68) and 30% (17/56) were low risk in Cohorts 1 and 2, respectively. In women aged 49-60, using an ultralow risk threshold of the 70-gene signature, 10% of tumors in Cohort 1 were ultralow risk compared with 30% for women with screen-detected cancers in Cohort 2. Older age and method of detection (screening) are associated with a higher likelihood of a biologically low risk tumor. In women over age 50, biologically low risk tumors are frequent and tools that classify risk may minimize overtreatment.

A diagnostic gene profile for molecular subtyping of breast cancer associated with treatment response

Classification of breast cancer into molecular subtypes maybe important for the proper selection of therapy, as tumors with seemingly similar histopathological features can have strikingly different clinical outcomes. Herein, we report the development of a molecular subtyping profile (BluePrint), that enables rationalization in patient selection for either chemotherapy or endocrine therapy prescription. An 80-Gene Molecular Subtyping Profile (BluePrint) was developed using 200 breast cancer patient specimens and confirmed on four independent validation cohorts (n = 784). Additionally, the profile was tested as a predictor of chemotherapy response in 133 breast cancer patients, treated with T/FAC neoadjuvant chemotherapy. BluePrint classification of a patient cohort that was treated with neoadjuvant chemotherapy (n = 133) shows improved distribution of pathological Complete Response (pCR), among molecular subgroups compared with local pathology: 56% of the patients had a pCR in the Basal-type subgroup, 3% in the MammaPrint Low-risk, Luminal-type subgroup, 11% in the MammaPrint High-risk, Luminal-type subgroup, and 50% in the HER2-type subgroup. The group of genes identifying Luminal-type breast cancer is highly enriched for genes having an Estrogen Receptor binding site proximal to the promoter-region, suggesting that these genes are direct targets of the Estrogen Receptor. Implementation of this profile may improve the clinical management of breast cancer patients, by enabling the selection of patients who are most likely to benefit from either chemotherapy or from endocrine therapy.

Comparison of MammaPrint and TargetPrint results with clinical parameters in German patients with early stage breast cancer

The 70-gene expression profile MammaPrint is a powerful prognostic indicator for disease outcome in breast cancer patients with improved prediction of recurrence risk compared to currently used guidelines. The microarray-based test TargetPrint further provides reliable, quantitative assessment of mRNA expression levels of estrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER2). This study was performed as a validation of MammaPrint and TargetPrint in an unselected German breast cancer population and was designed to determine the degree of concordance with currently applied clinical parameters. One hundred and forty cases of breast cancer stage I and II were classified as being low or high risk for distant metastasis using MammaPrint. Results were compared to current clinical risk classifications and adjuvant treatment management. Immunohistochemistry (IHC) and fluorescent in situ hybridization (FISH)/chromogenic in situ hybridization (CISH) assessments of ER, PR and HER2 were further compared with gene expression read-outs using TargetPrint. Thirty-two percent of patients (19/59) with a poor prognosis-signature identified via MammaPrint did not receive adjuvant systemic treatment apart from endocrine therapy and were potentially undertreated; whereas 42% (35/77) of patients with a good prognosis-signature received chemotherapy and were potentially overtreated. Comparison of microarray receptor results with IHC and FISH/CISH were concordant in 97% for ER; 86% for PR; and 94% for HER2. In this German study population, MammaPrint would have resulted in altered treatment advice for adjuvant systemic therapy in 40% of patients. Furthermore, TargetPrint presented high concordance for ER, PR and Her2 with IHC and FISH/CISH analysis.

Biological functions of the genes in the mammaprint breast cancer profile reflect the hallmarks of cancer

Background:

MammaPrint was developed as a diagnostic tool to predict risk of breast cancer metastasis using the expression of 70 genes. To better understand the tumor biology assessed by MammaPrint, we interpreted the biological functions of the 70-genes and showed how the genes reflect the six hallmarks of cancer as defined by Hanahan and Weinberg.

Results:

We used a bottom-up system biology approach to elucidate how the cellular processes reflected by the 70-genes work together to regulate tumor activities and progression. The biological functions of the genes were analyzed using literature research and several bioinformatics tools. Protein-protein interaction network analyses indicated that the 70-genes form highly interconnected networks and that their expression levels are regulated by key tumorigenesis related genes such as TP53, RB1, MYC, JUN and CDKN2A. The biological functions of the genes could be associated with the essential steps necessary for tumor progression and metastasis, and cover the six well-defined hallmarks of cancer, reflecting the acquired malignant characteristics of a cancer cell along with tumor progression and metastasis-related biological activities.

Conclusion:

Genes in the MammaPrint gene signature comprehensively measure the six hallmarks of cancer-related biology. This finding establishes a link between a molecular signature and the underlying molecular mechanisms of tumor cell progression and metastasis.

Gene expression signature to improve prognosis prediction of stage II and III colorectal cancer

PURPOSE

This study aims to develop a robust gene expression classifier that can predict disease relapse in patients with early-stage colorectal cancer (CRC).

PATIENTS AND METHODS

Fresh frozen tumor tissue from 188 patients with stage I to IV CRC undergoing surgery was analyzed using Agilent 44K oligonucleotide arrays. Median follow-up time was 65.1 months, and the majority of patients (83.6%) did not receive adjuvant chemotherapy. A nearest mean classifier was developed using a cross-validation procedure to score all genes for their association with 5-year distant metastasis-free survival.

RESULTS

An optimal set of 18 genes was identified and used to construct a prognostic classifier (ColoPrint). The signature was validated on an independent set of 206 samples from patients with stage I, II, and III CRC. The signature classified 60% of patients as low risk and 40% as high risk. Five-year relapse-free survival rates were 87.6% (95% CI, 81.5% to 93.7%) and 67.2% (95% CI, 55.4% to 79.0%) for low- and high-risk patients, respectively, with a hazard ratio (HR) of 2.5 (95% CI, 1.33 to 4.73; P = .005). In multivariate analysis, the signature remained one of the most significant prognostic factors, with an HR of 2.69 (95% CI, 1.41 to 5.14; P = .003). In patients with stage II CRC, the signature had an HR of 3.34 (P = .017) and was superior to American Society of Clinical Oncology criteria in assessing the risk of cancer recurrence without prescreening for microsatellite instability (MSI).

CONCLUSION

ColoPrint significantly improves the prognostic accuracy of pathologic factors and MSI in patients with stage II and III CRC and facilitates the identification of patients with stage II disease who may be safely managed without chemotherapy.

A gene expression profile for detection of sufficient tumour cells in breast tumour tissue: microarray diagnosis eligibility

Background

Microarray diagnostics of tumour samples is based on measurement of prognostic and/or predictive gene expression profiles. Typically, diagnostic profiles have been developed using bulk tumour samples with a sufficient amount of tumour cells (usually >50%). Consequentially, a diagnostic results depends on the minimal percentage of tumour cells within a sample. Currently, tumour cell percentage is assessed by conventional histopathological review. However, even for experienced pathologists, such scoring remains subjective and time consuming and can lead to ambiguous results.

Methods

In this study we investigated whether we could use transcriptional activity of a specific set of genes instead of histopathological review to identify samples with sufficient tumour cell content. Genome-wide gene expression measurements were used to develop a transcriptional gene profile that could accurately assess a sample's tumour cell percentage.

Results

Supervised analysis across 165 breast tumour samples resulted in the identification of a set of 13 genes which expression correlated with presence of tumour cells. The developed gene profile showed a high performance (AUC 0.92) for identification of samples that are suitable for microarray diagnostics. Validation on 238 additional breast tumour samples indicated a robust performance for correct classification with an overall accuracy of 91 percent and a kappa score of 0.63 (95%CI 0.47–0.73).

Conclusion

The developed 13-gene profile provides an objective tool for assessment whether a breast cancer sample contains sufficient tumour cells for microarray diagnostics. It will improve the efficiency and throughput for diagnostic gene expression profiling as it no longer requires histopathological analysis for initial tumour percentage scoring. Such profile will also be very use useful for assessment of tumour cell percentage in biopsies where conventional histopathology is difficult, such as fine needle aspirates.

Validation of 70-gene prognosis signature in node-negative breast cancer

PURPOSE

The 70-gene prognosis signature (van't Veer et al., Nature 415(6871):530-536, 2002) may improve the selection of lymph node-negative breast cancer patients for adjuvant systemic therapy. Optimal validation of prognostic classifiers is of great importance and we therefore wished to evaluate the prognostic value of the 70-gene prognosis signature in a series of relatively recently diagnosed lymph node negative breast cancer patients.

METHODS

We evaluated the 70-gene prognosis signature in an independent representative series of patients with invasive breast cancer (N = 123; <55 years; pT1-2N0; diagnosed between 1996 and 1999; median follow-up 5.8 years) by classifying these patients as having a good or poor prognosis signature. In addition, we updated the follow-up of the node-negative patients of the previously published validation-series (Van de Vijver et al., N Engl J Med 347(25):1999-2009, 2002; N = 151; median follow-up 10.2 years). The prognostic value of the 70-gene prognosis signature was compared with that of four commonly used clinicopathological risk indexes. The endpoints were distant metastasis (as first event) free percentage (DMFP) and overall survival (OS).

RESULTS

The 5-year OS was 82 +/- 5% in poor (48%) and 97 +/- 2% in good prognosis signature (52%) patients (HR 3.4; 95% CI 1.2-9.6; P = 0.021). The 5-years DMFP was 78 +/- 6% in poor and 98 +/- 2% in good prognosis signature patients (HR 5.7; 95% CI 1.6-20; P = 0.007). In the updated series (N = 151; 60% poor vs. 40% good), the 10-year OS was 51 +/- 5% and 94 +/- 3% (HR 10.7; 95% CI 3.9-30; P < 0.01), respectively. The DMFP was 50 +/- 6% in poor and 86 +/- 5% in good prognosis signature patients (HR 5.5; 95% CI 2.5-12; P < 0.01). In multivariate analysis, the prognosis signature was a strong independent prognostic factor in both series, outperforming the clinicopathological risk indexes.

CONCLUSION

The 70-gene prognosis signature is also an independent prognostic factor in node-negative breast cancer patients for women diagnosed in recent years.

Gene Expression Profiling to Identify the Histogenetic Origin of Metastatic Adenocarcinomas of Unknown Primary

Purpose

Patients with adenocarcinoma of unknown primary origin (ACUP) constitute approximately 4% of all malignancies. For effective treatment of these patients, it is considered optimal to identify the primary tumor origins. Currently, the success rate of the diagnostic work-up is only 20% to 30%. Our goal was to evaluate the contribution of gene expression profiling for routine clinical practice in patients with ACUP.

Patients and Methods

Formalin-fixed, paraffin-embedded (FFPE) samples were obtained from 84 patients with a known primary adenocarcinoma and from 38 patients with ACUP. An extensive immunohistochemical panel classified 16 of the patients with ACUP, whereas 22 patients remained unclassified for their histogenetic origin. Information about staging procedures and clinical follow-up were available in all patient cases. The expression data were analyzed in relation to clinicopathologic variables and immunohistochemical results.

Results

The gene expression–based assay classified the primary site correctly in 70 (83%) of 84 patient cases of primary and metastatic tumors of known origin, with good sensitivity for the majority of the tumor classes and relatively poor sensitivity for primary lung adenocarcinoma. Gene expression profiling identified 15 (94%) of 16 patients with initial ACUP who were classified by immunohistochemistry, and it made a valuable contribution to a potential site of origin in 14 of the 22 patients with ACUP.

Conclusion

The gene expression platform can classify correctly from FFPE samples the majority of tumors classes both in patients with known primary and in patients with ACUP. Therefore, gene expression profiling represents an additional analytic approach to assist with the histogenetic diagnosis of patients with ACUP.

Analysis of the MammaPrint breast cancer assay in a predominantly postmenopausal cohort

PURPOSE

Most node-negative breast cancer patients are older and postmenopausal and are increasingly being offered adjuvant chemotherapy despite their low overall risk of distant relapse. A molecular diagnostic test with high negative predictive value (NPV) for distant metastasis in this subgroup would spare many older breast cancer patients adjuvant treatment.

EXPERIMENTAL DESIGN

We determined the NPV and positive predictive value of the MammaPrint assay in breast cancer patients who were consecutively diagnosed and treated at the Massachusetts General Hospital between 1985 and 1997. Primary tumors from 100 patients with node-negative, invasive breast cancer (median age, 62.5 years; median follow-up, 11.3 years) were subjected to MammaPrint analysis and classified as being at either low or high risk for distant metastasis.

RESULTS

The MammaPrint 70-gene signature displayed excellent NPV as in previous studies, correctly identifying 100% of women at low risk for distant metastases at 5 years. However, this assay had a lower positive predictive value (12% at 5 years) than previously observed.

CONCLUSIONS

The MammaPrint assay was originally designed to identify younger breast cancer patients at low risk for distant metastasis, who might consequently be spared systemic treatment. We show here that the same signature has a very high NPV for distant recurrence after adjuvant treatment in older breast cancer patients.

PRAME expression and clinical outcome of breast cancer

The tumour antigen PReferentially expressed Antigen of MElanoma (PRAME) is expressed in a variety of malignancies, including breast cancer. We have analysed PRAME gene expression in relation to clinical outcome for 295 primary breast cancer patients. Kaplan–Meier survival curves show a correlation of PRAME expression levels with increased rates of distant metastases and decreased overall patient survival. This correlation existed both for the entire patient group (n=295) and for the subgroup of patients (n=185) who did not receive adjuvant chemotherapy. Multivariable analysis indicated that PRAME is an independent marker of shortened metastasis-free interval in patients who did not receive adjuvant chemotherapy. PRAME expression was associated with tumour grade and negative oestrogen receptor status. We conclude that PRAME expression is a prognostic marker for clinical outcome of breast cancer, independent of traditional clinicopathological markers.

Primary testicular diffuse large B-cell lymphomas have activated B-cell-like subtype characteristics

Diffuse large B-cell lymphomas (DLBCLs) constitute a heterogeneous group of lymphomas in which germinal centre B-cell-like and activated B-cell-like subtypes can be discerned based on pathology, clinical presentation, and gene expression patterns. Testicular DLBCLs form an immune-privileged site-related subgroup of DLBCLs with an unfavourable prognosis. In the present study, cDNA microarray analysis, immunohistochemistry for CD10, Bcl6 and MUM1, and somatic hypermutation analysis of the immunoglobulin heavy chain gene rearrangements were used to determine the subtype of primary testicular DLBCL. Immunohistochemistry revealed 14/22 testicular DLBCLs with an activated B-cell-like immunophenotype and 8/22 with an ambiguous immunophenotype co-expressing CD10 and high levels of MUM1. cDNA microarray analysis of these 22 and four additional cases showed a uniform activated B-cell-like gene expression pattern for both immunophenotypes. Somatic hypermutation analysis of immunoglobulin heavy chain genes showed a very high mutation load in seven cases tested, but intraclonal heterogeneity was found at low level in only one of these cases. It is concluded that primary testicular DLBCLs have uniform activated B-cell-like subtype characteristics despite a number of cases showing an ambiguous immunophenotype.

Robust interlaboratory reproducibility of a gene expression signature measurement consistent with the needs of a new generation of diagnostic tools

Background

The increasing use of DNA microarrays in biomedical research, toxicogenomics, pharmaceutical development, and diagnostics has focused attention on the reproducibility and reliability of microarray measurements. While the reproducibility of microarray gene expression measurements has been the subject of several recent reports, there is still a need for systematic investigation into what factors most contribute to variability of measured expression levels observed among different laboratories and different experimenters.

Results

We report the results of an interlaboratory comparison of gene expression array measurements on the same microarray platform, in which the RNA amplification and labeling, hybridization and wash, and slide scanning were each individually varied. Identical input RNA was used for all experiments. While some sources of variation have measurable influence on individual microarray signals, they showed very low influence on sample-to-reference ratios based on averaged triplicate measurements in the two-color experiments. RNA labeling was the largest contributor to interlaboratory variation.

Conclusion

Despite this variation, measurement of one particular breast cancer gene expression signature in three different laboratories was found to be highly robust, showing a high intralaboratory and interlaboratory reproducibility when using strictly controlled standard operating procedures.

Gene-expression and immunohistochemical study of specific T-cell subsets and accessory cell types in the transformation and prognosis of follicular lymphoma

PURPOSE

Despite the generally favorable clinical course in follicular lymphoma (FL), a minority of patients have a poor prognosis-with death within 3 years of diagnosis-most often due to transformation to aggressive disease.

PATIENTS AND METHODS

In this study, we analyzed the potential of predicting early transformation on the basis of gene expression and immunologic parameters in FL biopsy samples taken at diagnosis.

RESULTS

At the gene-expression level, FL is a highly uniform disease at the time of diagnosis, precluding the detection of sufficiently validated prognostic gene-expression profiles suitable for a clinical setting. Combinations of differentially expressed genes indicate that immunologic mechanisms play a differential role in the risk of early transformation. Using immunohistochemistry for specific cell populations, the spatial distribution to neoplastic follicles and the activation of CD4-positive T-helper cells (P = .002) and specifically T-helper 1 (P = .004) were shown to be highly discriminatory to predict early transformation. A role for functional modulation of follicular dendritic cells could also be supported (P = .04). Other cell populations, including CD68-positive macrophages and regulatory T cells, were not differentially present.

CONCLUSION

These results support the identification of FL as an immunologically functional disease in which an interaction of the tumor cells and the functional composition of the microenvironment determines the clinical behavior.

Converting a breast cancer microarray signature into a high-throughput diagnostic test

BACKGROUND

A 70-gene tumor expression profile was established as a powerful predictor of disease outcome in young breast cancer patients. This profile, however, was generated on microarrays containing 25,000 60-mer oligonucleotides that are not designed for processing of many samples on a routine basis.

RESULTS

To facilitate its use in a diagnostic setting, the 70-gene prognosis profile was translated into a customized microarray (MammaPrint) containing a reduced set of 1,900 probes suitable for high throughput processing. RNA of 162 patient samples from two previous studies was subjected to hybridization to this custom array to validate the prognostic value. Classification results obtained from the original analysis were then compared to those generated using the algorithms based on the custom microarray and showed an extremely high correlation of prognosis prediction between the original data and those generated using the custom mini-array (p < 0.0001).

CONCLUSION

In this report we demonstrate for the first time that microarray technology can be used as a reliable diagnostic tool. The data clearly demonstrate the reproducibility and robustness of the small custom-made microarray. The array is therefore an excellent tool to predict outcome of disease in breast cancer patients.

Validation and clinical utility of a 70-gene prognostic signature for women with node-negative breast cancer

BACKGROUND

A 70-gene signature was previously shown to have prognostic value in patients with node-negative breast cancer. Our goal was to validate the signature in an independent group of patients.

METHODS

Patients (n = 307, with 137 events after a median follow-up of 13.6 years) from five European centers were divided into high- and low-risk groups based on the gene signature classification and on clinical risk classifications. Patients were assigned to the gene signature low-risk group if their 5-year distant metastasis-free survival probability as estimated by the gene signature was greater than 90%. Patients were assigned to the clinicopathologic low-risk group if their 10-year survival probability, as estimated by Adjuvant! software, was greater than 88% (for estrogen receptor [ER]-positive patients) or 92% (for ER-negative patients). Hazard ratios (HRs) were estimated to compare time to distant metastases, disease-free survival, and overall survival in high- versus low-risk groups.

RESULTS

The 70-gene signature outperformed the clinicopathologic risk assessment in predicting all endpoints. For time to distant metastases, the gene signature yielded HR = 2.32 (95% confidence interval [CI] = 1.35 to 4.00) without adjustment for clinical risk and hazard ratios ranging from 2.13 to 2.15 after adjustment for various estimates of clinical risk; clinicopathologic risk using Adjuvant! software yielded an unadjusted HR = 1.68 (95% CI = 0.92 to 3.07). For overall survival, the gene signature yielded an unadjusted HR = 2.79 (95% CI = 1.60 to 4.87) and adjusted hazard ratios ranging from 2.63 to 2.89; clinicopathologic risk yielded an unadjusted HR = 1.67 (95% CI = 0.93 to 2.98). For patients in the gene signature high-risk group, 10-year overall survival was 0.69 for patients in both the low- and high-clinical risk groups; for patients in the gene signature low-risk group, the 10-year survival rates were 0.88 and 0.89, respectively.

CONCLUSIONS

The 70-gene signature adds independent prognostic information to clinicopathologic risk assessment for patients with early breast cancer.

Mechanisms and effects of loss of human leukocyte antigen class II expression in immune-privileged site-associated B-cell lymphoma

PURPOSE AND EXPERIMENTAL DESIGN

Loss of human leukocyte antigen (HLA) expression on tumor cells is frequent in diffuse large B-cell lymphoma (DLBCL) arising in immune-privileged sites, such as the testis and central nervous system, and is associated with small homozygous deletions of HLA-DQ/HLA-DR and larger hemizygous deletions of the MHC region. To better understand the significance of down-regulation of HLA class II expression in relation to the homozygous and hemizygous deletions, we analyzed global gene expression patterns in a series of 26 testicular DLBCL after characterization of these deletions.

RESULTS

Low levels of HLA-DR mRNA in whole testicular DLBCL samples were associated with a strong down-regulation of numerous immune-related genes specific for T cells, macrophages, antigen presentation and processing, lymphocyte activation, chemokines and chemokine receptors, and the complement system. The number of CD3+ tumor-infiltrating T cells was also significantly lower in low expressors of HLA-DR mRNA. Interestingly, hemizygous and homozygous deletions in the MHC region did not have any additional effect on global gene expression.

CONCLUSION

In conclusion, we found that loss of HLA class II mRNA expression in testicular DLBCL is associated with a significant change in global gene expression patterns. This effect is independent of the mechanism causing the down-regulation of HLA class II genes in the lymphoma cells.

No common denominator for breast cancer lymph node metastasis

The axillary lymph node status is the most powerful prognostic factor for breast cancer patients to date. The molecular mechanisms that control lymph node metastasis, however, remain poorly understood. To define patterns of genes or gene regulatory pathways that drive breast cancer lymph node metastasis, we compared the gene expression profiles of 15 primary breast carcinomas and their matching lymph node metastases using microarrays. In general, primary breast carcinomas and lymph node metastases do not differ at the transcriptional level by a common subset of genes. No classifier or single gene discriminating the group of primary tumours from those of the lymph node metastases could be identified. Also, in a series of 295 breast tumours, no classifier predicting lymph node metastasis could be developed. However, subtle differences in the expression of genes involved in extracellular-matrix organisation and growth factor signalling are detected in individual pairs of matching primary and metastatic tumours. Surprisingly, however, different sets of these genes are either up- or downregulated in lymph node metastases. Our data suggest that breast carcinomas do not use a shared gene set to accomplish lymph node metastasis.

Molecular portraits and 70-gene prognosis signature are preserved throughout the metastatic process of breast cancer

Microarray analysis has been shown to improve risk stratification of breast cancer. Breast tumors analyzed by hierarchical clustering of expression patterns of "intrinsic" genes have been reported to subdivide into at least four molecular subtypes that are associated with distinct patient outcomes. Using a supervised method, a 70-gene expression profile has been identified that predicts the later appearance or absence of clinical metastasis in young breast cancer patients. Here, we show that distant metastases display both the same molecular breast cancer subtype as well as the 70-gene prognosis signature as their primary tumors. Our results suggest that the capacity to metastasize is an inherent feature of most breast cancers. Furthermore, our data imply that poor prognosis breast carcinomas classified either by the intrinsic gene set or the 70 prognosis genes represent distinct disease entities that seem sustained throughout the metastatic process.

Gene expression profiling in follicular lymphoma to assess clinical aggressiveness and to guide the choice of treatment

Follicular lymphoma (FL) is a disease characterized by a long clinical course marked by frequent relapses that vary in clinical aggressiveness over time. Therefore, the main dilemma at each relapse is the choice for the most effective treatment for optimal disease control and failure-free survival while at the same time avoiding overtreatment and harmful side effects. The selection for more aggressive treatment is currently based on histologic grading and clinical criteria; however, in up to 30% of all cases these methods prove to be insufficient. Using supervised classification on a training set of paired samples from patients who experienced either an indolent or aggressive disease course, a gene expression profile of 81 genes was established that could, with an accuracy of 100%, distinguish low-grade from high-grade disease. This profile accurately classified 93% of the FL samples in an independent validation set. Most important, in a third series of FL cases where histologic grading was ambiguous, precluding meaningful morphologic guidance, the 81-gene profile shows a classification accuracy of 94%. The FL stratification profile is a more reliable marker of clinical behavior than the currently used histologic grading and clinical criteria and may provide an important alternative to guide the choice of therapy in patients with FL both at presentation and at relapse.