Design and analysis of ChIP-Seq experiments for nuclear factor I DNA-binding proteins

SUMMARY : Eukaryotic DNA interacts with the nuclear proteins using non-covalent ionic interactions. Proteins can recognize specific nucleotide sequences based on the sterical interactions with the DNA and these specific protein-DNA interactions are the basis for many nuclear processes, e.g. gene transcription, chromosomal replication, and recombination. New technology termed ChIP-Seq has been recently developed for the analysis of protein-DNA interactions on a whole genome scale and it is based on immunoprecipitation of chromatin and high-throughput DNA sequencing procedure. ChIP-Seq is a novel technique with a great potential to replace older techniques for mapping of protein-DNA interactions. In this thesis, we bring some new insights into the ChIP-Seq data analysis. First, we point out to some common and so far unknown artifacts of the method. Sequence tag distribution in the genome does not follow uniform distribution and we have found extreme hot-spots of tag accumulation over specific loci in the human and mouse genomes. These artifactual sequence tags accumulations will create false peaks in every ChIP-Seq dataset and we propose different filtering methods to reduce the number of false positives. Next, we propose random sampling as a powerful analytical tool in the ChIP-Seq data analysis that could be used to infer biological knowledge from the massive ChIP-Seq datasets. We created unbiased random sampling algorithm and we used this methodology to reveal some of the important biological properties of Nuclear Factor I DNA binding proteins. Finally, by analyzing the ChIP-Seq data in detail, we revealed that Nuclear Factor I transcription factors mainly act as activators of transcription, and that they are associated with specific chromatin modifications that are markers of open chromatin. We speculate that NFI factors only interact with the DNA wrapped around the nucleosome. We also found multiple loci that indicate possible chromatin barrier activity of NFI proteins, which could suggest the use of NFI binding sequences as chromatin insulators in biotechnology applications. RESUME : L'ADN des eucaryotes interagit avec les protéines nucléaires par des interactions noncovalentes ioniques. Les protéines peuvent reconnaître les séquences nucléotidiques spécifiques basées sur l'interaction stérique avec l'ADN, et des interactions spécifiques contrôlent de nombreux processus nucléaire, p.ex. transcription du gène, la réplication chromosomique, et la recombinaison. Une nouvelle technologie appelée ChIP-Seq a été récemment développée pour l'analyse des interactions protéine-ADN à l'échelle du génome entier et cette approche est basée sur l'immuno-précipitation de la chromatine et sur la procédure de séquençage de l'ADN à haut débit. La nouvelle approche ChIP-Seq a donc un fort potentiel pour remplacer les anciennes techniques de cartographie des interactions protéine-ADN. Dans cette thèse, nous apportons de nouvelles perspectives dans l'analyse des données ChIP-Seq. Tout d'abord, nous avons identifié des artefacts très communs associés à cette méthode qui étaient jusqu'à présent insoupçonnés. La distribution des séquences dans le génome ne suit pas une distribution uniforme et nous avons constaté des positions extrêmes d'accumulation de séquence à des régions spécifiques, des génomes humains et de la souris. Ces accumulations des séquences artéfactuelles créera de faux pics dans toutes les données ChIP-Seq, et nous proposons différentes méthodes de filtrage pour réduire le nombre de faux positifs. Ensuite, nous proposons un nouvel échantillonnage aléatoire comme un outil puissant d'analyse des données ChIP-Seq, ce qui pourraient augmenter l'acquisition de connaissances biologiques à partir des données ChIP-Seq. Nous avons créé un algorithme d'échantillonnage aléatoire et nous avons utilisé cette méthode pour révéler certaines des propriétés biologiques importantes de protéines liant à l'ADN nommés Facteur Nucléaire I (NFI). Enfin, en analysant en détail les données de ChIP-Seq pour la famille de facteurs de transcription nommés Facteur Nucléaire I, nous avons révélé que ces protéines agissent principalement comme des activateurs de transcription, et qu'elles sont associées à des modifications de la chromatine spécifiques qui sont des marqueurs de la chromatine ouverte. Nous pensons que lés facteurs NFI interagir uniquement avec l'ADN enroulé autour du nucléosome. Nous avons également constaté plusieurs régions génomiques qui indiquent une éventuelle activité de barrière chromatinienne des protéines NFI, ce qui pourrait suggérer l'utilisation de séquences de liaison NFI comme séquences isolatrices dans des applications de la biotechnologie.

Molecular pathology of colorectal cancer.

Colorectal cancer (CRC) is one of the most intensively studied cancer types, partly because of its high prevalence but also because of the existence of its precursor lesions, tubular or villous adenomas, and more recently (sessile) serrated adenomas, which can be detected endoscopically and removed. The morphological steps in the adenoma-carcinoma sequence have been elucidated at a molecular level, which has been facilitated by identification of the genes responsible for familial intestinal cancer. However, apart from early detection of familial forms of CRC and its use in genetic counseling, until recently such detailed molecular knowledge has had little impact on clinical management of the disease. This has dramatically changed in the last decade. With drugs specifically targeting the epidermal growth factor receptor (EGFR) having been shown effective in CRC, mechanisms responsible for resistance have been explored. The finding that KRAS mutated cancers do not respond to anti-EGFR treatment has had a profound impact on clinical management and on molecular diagnostics of CRC. Additional genetic tests for mutations in NRAS, BRAF and PIK3CA contribute to determining who to treat, and others will follow. New therapies effective in patients with advanced CRC are under investigation. Remaining burning questions for optimal management are which patients will relapse after resection of the primary tumor and which patients will respond to the standard 5FU-oxaliplatin adjuvant treatment regimen. Predictive tests to address these issues are eagerly awaited. New classifications of CRC, based on molecular parameters, are emerging, and we will be confronted with new subtypes of CRC, for which the definition is based on combinations of gene expression patterns, chromosomal alterations, gene mutations and epigenetic characteristics. This will be instrumental in designing new approaches for therapy but will also be translated into molecular diagnostics. Both will contribute to improved clinical management of CRC.

Repeated injections of 131I-rituximab show patient-specific stable biodistribution and tissue kinetics.

PURPOSE: It is generally assumed that the biodistribution and pharmacokinetics of radiolabelled antibodies remain similar between dosimetric and therapeutic injections in radioimmunotherapy. However, circulation half-lives of unlabelled rituximab have been reported to increase progressively after the weekly injections of standard therapy doses. The aim of this study was to evaluate the evolution of the pharmacokinetics of repeated 131I-rituximab injections during treatment with unlabelled rituximab in patients with non-Hodgkin's lymphoma (NHL). METHODS: Patients received standard weekly therapy with rituximab (375 mg/m2) for 4 weeks and a fifth injection at 7 or 8 weeks. Each patient had three additional injections of 185 MBq 131I-rituximab in either treatment weeks 1, 3 and 7 (two patients) or weeks 2, 4 and 8 (two patients). The 12 radiolabelled antibody injections were followed by three whole-body (WB) scintigraphic studies during 1 week and blood sampling on the same occasions. Additional WB scans were performed after 2 and 4 weeks post 131I-rituximab injection prior to the second and third injections, respectively. RESULTS: A single exponential radioactivity decrease for WB, liver, spleen, kidneys and heart was observed. Biodistribution and half-lives were patient specific, and without significant change after the second or third injection compared with the first one. Blood T(1/2)beta, calculated from the sequential blood samples and fitted to a bi-exponential curve, was similar to the T(1/2) of heart and liver but shorter than that of WB and kidneys. Effective radiation dose calculated from attenuation-corrected WB scans and blood using Mirdose3.1 was 0.53+0.05 mSv/MBq (range 0.48-0.59 mSv/MBq). Radiation dose was highest for spleen and kidneys, followed by heart and liver. CONCLUSION: These results show that the biodistribution and tissue kinetics of 131I-rituximab, while specific to each patient, remained constant during unlabelled antibody therapy. RIT radiation doses can therefore be reliably extrapolated from a preceding dosimetry study.

Hybridisation and diversification in the adaptive radiation of clownfishes.

BackgroundThe importance of hybridisation during species diversification has long been debated among evolutionary biologists. It is increasingly recognised that hybridisation events occurred during the evolutionary history of numerous species, especially during the early stages of adaptive radiation. We study the effect of hybridisation on diversification in the clownfishes, a clade of coral reef fish that diversified through an adaptive radiation process. While two species of clownfish are likely to have been described from hybrid specimens, the occurrence and effect of hybridisation on the clade diversification is yet unknown.ResultsWe generate sequences of three mitochondrial genes to complete an existing dataset of nuclear sequences and document cytonuclear discordance at a node, which shows a drastic increase of diversification rate. Then, using a tree-based jack-knife method, we identify clownfish species likely stemming from hybridisation events. Finally, we use molecular cloning and identify the putative parental species of four clownfish specimens that display the morphological characteristics of hybrids.ConclusionsOur results show that consistently with the syngameon hypothesis, hybridisation events are linked with a burst of diversification in the clownfishes. Moreover, several recently diverged clownfish lineages likely originated through hybridisation, which indicates that diversification, catalysed by hybridisation events, may still be happening.

Effect of Platinum-Based Chemoradiotherapy on Cellular Proliferation in Bone Marrow and Spleen, Estimated by 18F-FLT PET/CT in Patients with Locally Advanced Non-Small Cell Lung Cancer.

Historically, it has been difficult to monitor the acute impact of anticancer therapies on hematopoietic organs on a whole-body scale. Deeper understanding of the effect of treatments on bone marrow would be of great potential value in the rational design of intensive treatment regimens. 3'-deoxy-3'-(18)F-fluorothymidine ((18)F-FLT) is a functional radiotracer used to study cellular proliferation. It is trapped in cells in proportion to thymidine-kinase 1 enzyme expression, which is upregulated during DNA synthesis. This study investigates the potential of (18)F-FLT to monitor acute effects of chemotherapy on cellular proliferation and its recovery in bone marrow, spleen, and liver during treatment with 2 different chemotherapy regimens.

Comparison of copper-67- and iodine-125-labeled anti-CEA monoclonal antibody biodistribution in patients with colorectal tumors.

Copper-67 has comparable beta-particle emissions to that of 131I, but it displays more favorable gamma emission characteristics for application in radioimmunotherapy (RIT). This study investigates the potential of 67Cu-labeled monoclonal antibody (MAb) 35 for RIT of colorectal carcinoma. METHODS: Biokinetics of simultaneously injected 67Cu- and 125I-labeled MAb35 were studied in six patients scheduled for surgery of primary colorectal cancer. RESULTS: Whole-body clearance (T 1/2) of 67Cu, estimated from sequential anterior and posterior whole-body scans and corrected for decay of 67Cu, was 41 hr. Serum clearance of 67Cu was faster (27.41 hr) than that of 125I (38.33 hr). Mean tumor uptake of the 67Cu-labeled compound (0.0133% ID/g) exceeded that of 125I (0.0095% ID/g), and tumor-to-blood ratios were higher for 67Cu than for 125I, with averages of 6.07 and 2.41, respectively. The average 67Cu/125I ratio was 1.9 for tumor uptake, 0.7 for blood and 2.6 for tumor-to-blood ratios. Nonspecific liver uptake of 67Cu as calculated from whole-body scans was high in four patients, up to 25% of residual whole-body activity at 48 hr, but did not increase with time. We also observed some nonspecific bowel activity, as well as moderate to high uptake in benign polyps. CONCLUSION: Copper-67-labeled MAb35 is more favorable than its radioiodine-labeled counterpart for RIT of colorectal carcinoma due to higher tumor-to-blood ratios, but the problem of nonspecific liver and bowel uptake must first be overcome. The absolute accumulation of activity in tumor remains low, however, so the probability of cure with this compound alone is questionable. The use of 67Cu as one component of a multimodality adjuvant treatment seems to remain the most appropriate application for RIT.

Fluorescence imaging reveals the nuclear behavior of peroxisome proliferator-activated receptor/retinoid X receptor heterodimers in the absence and presence of ligand.

In a global approach combining fluorescence recovery after photobleaching (FRAP), fluorescence correlation spectroscopy (FCS), and fluorescence resonance energy transfer (FRET), we address the behavior in living cells of the peroxisome proliferator-activated receptors (PPARs), a family of nuclear receptors involved in lipid and glucose metabolism, inflammation control, and wound healing. We first demonstrate that unlike several other nuclear receptors, PPARs do not form speckles upon ligand activation. The subnuclear structures that may be observed under some experimental conditions result from overexpression of the protein and our immunolabeling experiments suggest that these structures are subjected to degradation by the proteasome. Interestingly and in contrast to a general assumption, PPARs readily heterodimerize with retinoid X receptor (RXR) in the absence of ligand in living cells. PPAR diffusion coefficients indicate that all the receptors are engaged in complexes of very high molecular masses and/or interact with relatively immobile nuclear components. PPARs are not immobilized by ligand binding. However, they exhibit a ligand-induced reduction of mobility, probably due to enhanced interactions with cofactors and/or chromatin. Our study draws attention to the limitations and pitfalls of fluorescent chimera imaging and demonstrates the usefulness of the combination of FCS, FRAP, and FRET to assess the behavior of nuclear receptors and their mode of action in living cells.

N-Terminal Modifications Improve the Receptor Affinity and Pharmacokinetics of Radiolabeled Peptidic Gastrin-Releasing Peptide Receptor Antagonists: Examples of 68Ga- and 64Cu-Labeled Peptides for PET Imaging.

Gastrin-releasing peptide receptors (GRPrs) are overexpressed on a variety of human cancers, providing the opportunity for peptide receptor targeting via radiolabeled bombesin-based peptides. As part of our ongoing investigations into the development of improved GRPr antagonists, this study aimed at verifying whether and how N-terminal modulations improve the affinity and pharmacokinetics of radiolabeled GRPr antagonists. METHODS: The potent GRPr antagonist MJ9, Pip-d-Phe-Gln-Trp-Ala-Val-Gly-His-Sta-Leu-NH2 (Pip, 4-amino-1-carboxymethyl-piperidine), was conjugated to 1,4,7-triazacyclononane, 1-glutaric acid-4,7 acetic acid (NODAGA), and 1,4,7-triazacyclononane-1,4,7-triacetic acid (NOTA) and radiolabeled with (68)Ga and (64)Cu. The GRPr affinity of the corresponding metalloconjugates was determined using (125)I-Tyr(4)-BN as a radioligand. The labeling efficiency of (68)Ga(3+) was compared between NODAGA-MJ9 and NOTA-MJ9 in acetate buffer, at room temperature and at 95°C. The (68)Ga and (64)Cu conjugates were further evaluated in vivo in PC3 tumor xenografts by biodistribution and PET imaging studies. RESULTS: The half maximum inhibitory concentrations of all the metalloconjugates are in the high picomolar-low nanomolar range, and these are the most affine-radiolabeled GRPr antagonists we have studied so far in our laboratory. NODAGA-MJ9 incorporates (68)Ga(3+) nearly quantitatively (>98%) at room temperature within 10 min and at much lower peptide concentrations (1.4 × 10(-6) M) than NOTA-MJ9, for which the labeling yield was approximately 45% under the same conditions and increased to 75% at 95°C for 5 min. Biodistribution studies showed high and specific tumor uptake, with a maximum of 23.3 ± 2.0 percentage injected activity per gram of tissue (%IA/g) for (68)Ga-NOTA-MJ9 and 16.7 ± 2.0 %IA/g for (68)Ga-NODAGA-MJ9 at 1 h after injection. The acquisition of PET images with the (64)Cu-MJ9 conjugates at later time points clearly showed the efficient clearance of the accumulated activity from the background already at 4 h after injection, whereas tumor uptake still remained high. The high pancreas uptake for all radiotracers at 1 h after injection was rapidly washed out, resulting in an increased tumor-to-pancreas ratio at later time points. CONCLUSION: We have developed 2 GRPr antagonistic radioligands, which are improved in terms of binding affinity and overall biodistribution profile. Their promising in vivo pharmacokinetic performance may contribute to the improvement of the diagnostic imaging of tumors overexpressing GRPr.

Differential expression of THOC1 and ALY mRNP biogenesis/export factors in human cancers

One key step in gene expression is the biogenesis of mRNA ribonucleoparticle complexes (mRNPs). Formation of the mRNP requires the participation of a number of conserved factors such as the THO complex. THO interacts physically and functionally with the Sub2/UAP56 RNA-dependent ATPase, and the Yra1/REF1/ALY RNA-binding protein linking transcription, mRNA export and genome integrity. Given the link between genome instability and cancer, we have performed a comparative analysis of the expression patterns of THOC1, a THO complex subunit, and ALY in tumor samples.

On the use of principal components in contemporany population genetics: a case study

In human Population Genetics, routine applications of principal component techniques are oftenrequired. Population biologists make widespread use of certain discrete classifications of humansamples into haplotypes, the monophyletic units of phylogenetic trees constructed from severalsingle nucleotide bimorphisms hierarchically ordered. Compositional frequencies of the haplotypesare recorded within the different samples. Principal component techniques are then required as adimension-reducing strategy to bring the dimension of the problem to a manageable level, say two,to allow for graphical analysis.Population biologists at large are not aware of the special features of compositional data and normally make use of the crude covariance of compositional relative frequencies to construct principalcomponents. In this short note we present our experience with using traditional linear principalcomponents or compositional principal components based on logratios, with reference to a specificdataset

Bone Mineral Density and Serum Levels of Soluble Tumor Necrosis Factors, Estradiol, and Osteoprotegerin in Postmenopausal Women with Cirrhosis after Viral Hepatitis

CONTEXT: Cirrhosis after viral hepatitis has been identified as a risk factor for osteoporosis in men. However, in postmenopausal women, most studies have evaluated the effect of primary biliary cirrhosis, but little is known about the effect of viral cirrhosis on bone mass [bone mineral density (BMD)] and bone metabolism. OBJECTIVE: Our objective was to assess the effect of viral cirrhosis on BMD and bone metabolism in postmenopausal women. DESIGN: We conducted a cross-sectional descriptive study. SETTING AND PATIENTS: We studied 84 postmenopausal female outpatients with viral cirrhosis and 96 healthy postmenopausal women from the general community. BMD was measured by dual-energy x-ray absorptiometry at lumbar spine (LS) and femoral neck (FN). RESULTS: The percentage with osteoporosis did not significantly differ between patients (LS, 43.1%; FN, 32.2%) and controls (LS, 41.2%; FN, 29.4%), and there was no difference in BMD (z-score) between groups. Serum concentrations of soluble TNF receptors, estradiol, and osteoprotegerin (OPG) were significantly higher in patients vs. controls (P < 0.001, P < 0.05, and P < 0.05, respectively). No significant difference was observed in urinary deoxypyridinoline. Serum OPG levels were positively correlated with soluble TNF receptors (r = 0.35; P < 0.02) and deoxypyridinoline (r = 0.37; P < 0.05). CONCLUSIONS: This study shows that bone mass and bone resorption rates do not differ between postmenopausal women with viral cirrhosis and healthy postmenopausal controls and suggests that viral cirrhosis does not appear to increase the risk of osteoporosis in these women. High serum estradiol and OPG concentrations may contribute to preventing the bone loss associated with viral cirrhosis in postmenopausal women.

Expression and localization of PPARs in the rat ovary during follicular development and the periovulatory period.

PPARs are a family of nuclear hormone receptors involved in various processes that could influence ovarian function. We investigated the cellular localization and expression of PPARs during follicular development in ovarian tissue collected from rats 0, 6, 12, 24, and 48 h post-PMSG. A second group of animals received human CG (hCG) 48 h post-PMSG. Their ovaries were removed 0, 4, 8, 12, and 24 h post-hCG to study the periovulatory period. mRNAs corresponding to the PPAR isotypes (alpha, delta, and gamma) were localized by in situ hybridization. Changes in the levels of mRNA for the PPARs were determined by ribonuclease protection assays. PPAR gamma mRNA was localized primarily to granulosa cells, and levels of expression did not change during follicular development. Four hours post-hCG, levels of mRNA for PPAR gamma decreased (P &lt; 0.05) but not uniformly in all follicles. At 24 h post-hCG, levels of PPAR gamma mRNA were reduced 64%, but some follicles maintained high expression. In contrast, mRNAs for PPAR alpha and delta were located primarily in theca and stroma, and their levels did not change during the intervals studied. To investigate the physiologic significance of PPAR gamma in the ovary, granulosa cells from PMSG-primed rats were cultured for 48 h with prostaglandin J(2) (PGJ(2)) and ciglitazone, PPAR gamma activators. Both compounds increased progesterone and E2 secretion (P &lt; 0.05). These data suggest that PPAR gamma is involved in follicular development, has a negative influence on the luteinization of granulosa cells, and/or regulates the periovulatory shift in steroid production. The more general and steady expression of PPARs alpha and delta indicate that they may play a role in basal ovarian function.

Prevalence of symptomatic and silent stress-induced perfusion defects in diabetic patients with suspected coronary artery disease referred for myocardial perfusion scintigraphy.

PURPOSE: Silent myocardial ischaemia--as evaluated by stress-induced perfusion defects on myocardial perfusion scintigraphy (MPS) in patients without a history of chest pain--is frequent in diabetes and is associated with increased rates of cardiovascular events. Its prevalence has been determined in asymptomatic diabetic patients, but remains largely unknown in diabetic patients with suspected coronary artery disease (CAD) in the clinical setting. In this study we therefore sought (a) to determine the prevalence of symptomatic and silent perfusion defects in diabetic patients with suspected CAD and (b) to characterise the eventual predictors of abnormal perfusion. METHODS: The patient population comprised 133 consecutive diabetic patients with suspected CAD who had been referred for MPS. Studies were performed with exercise (41%) or pharmacological stress testing (1-day protocol, (99m)Tc-sestamibi, 201Tl or both). We used semi-quantitative analysis (20-segment polar maps) to derive the summed stress score (SSS) and the summed difference score (SDS). RESULTS: Abnormal MPS (SSS> or =4) was observed in 49 (37%) patients (SSS=4.9+/-8.4, SDS=2.4+/-4.7), reversible perfusion defects (SDS> or =2) in 40 (30%) patients [SSS=13.3+/-10.9; SDS=8.0+/-5.6; 20% moderate to severe (SDS>4), 7% multivessel] and fixed defects in 21 (16%) patients. Results were comparable between patients with and patients without a history of chest pain. Of 75 patients without a history of chest pain, 23 (31%, 95% CI=21-42%) presented reversible defects (SSS=13.9+/-11.3; SDS=7.4+/-1.2), indicative of silent ischaemia. Reversible defects were associated with inducible ST segment depression during MPS stress [odds ratio (OR)=3.2, p<0.01). Fixed defects were associated with erectile dysfunction in males (OR=3.7, p=0.02) and lower aspirin use (OR=0.25, p=0.02). CONCLUSION: Silent stress-induced perfusion defects occurred in 31% of the patients, a rate similar to that in patients with a history of chest pain. MPS could identify these patients with a potentially increased risk of cardiovascular events.

Pattern of recurrence of early breast cancer is different according to intrinsic subtype and proliferation index.

INTRODUCTION Recurrence risk in breast cancer varies throughout the follow-up time. We examined if these changes are related to the level of expression of the proliferation pathway and intrinsic subtypes. METHODS Expression of estrogen and progesterone receptor, Ki-67, human epidermal growth factor receptor 2 (HER2), epidermal growth factor receptor (EGFR) and cytokeratin 5/6 (CK 5/6) was performed on tissue-microarrays constructed from a large and uniformly managed series of early breast cancer patients (N = 1,249). Subtype definitions by four biomarkers were as follows: luminal A (ER + and/or PR+, HER2-, Ki-67 <14), luminal B (ER + and/or PR+, HER2-, Ki-67 ≥14), HER2-enriched (any ER, any PR, HER2+, any Ki-67), triple-negative (ER-, PR-, HER2-, any Ki-67). Subtype definitions by six biomarkers were as follows: luminal A (ER + and/or PR+, HER2-, Ki-67 <14, any CK 5/6, any EGFR), luminal B (ER + and/or PR+, HER2-, Ki-67 ≥14, any CK 5/6, any EGFR), HER2-enriched (ER-, PR-, HER2+, any Ki-67, any CK 5/6, any EGFR), Luminal-HER2 (ER + and/or PR+, HER2+, any Ki-67, any CK 5/6, any EGFR), Basal-like (ER-, PR-, HER2-, any Ki-67, CK5/6+ and/or EGFR+), triple-negative nonbasal (ER-, PR-, HER2-, any Ki-67, CK 5/6-, EGFR-). Each four- or six-marker defined intrinsic subtype was divided in two groups, with Ki-67 <14% or with Ki-67 ≥14%. Recurrence hazard rate function was determined for each intrinsic subtype as a whole and according to Ki-67 value. RESULTS Luminal A displayed a slow risk increase, reaching its maximum after three years and then remained steady. Luminal B presented most of its relapses during the first five years. HER2-enriched tumors show a peak of recurrence nearly twenty months post-surgery, with a greater risk in Ki-67 ≥14%. However a second peak occurred at 72 months but the risk magnitude was greater in Ki-67 <14%. Triple negative tumors with low proliferation rate display a smooth risk curve, but with Ki-67 ≥14% show sharp peak at nearly 18 months. CONCLUSIONS Each intrinsic subtype has a particular pattern of relapses over time which change depending on the level of activation of the proliferation pathway assessed by Ki-67. These findings could have clinical implications both on adjuvant treatment trial design and on the recommendations concerning the surveillance of patients.