WO2015162596A1

WO2015162596A1 - Use of immune diversity as a predictive marker for identifying patients likely to respond to an anti-ctla4 treatment

Info

Publication number: WO2015162596A1
Application number: PCT/IB2015/053007
Authority: WO
Inventors: Nicolas Pasqual; Manuel Manuarii; Anaïs Courtier; Jean-François MOURET; Sébastien WEISBUCH
Original assignee: Immunid
Priority date: 2014-04-25
Filing date: 2015-04-24
Publication date: 2015-10-29

Abstract

The present invention pertains to the field of cancer therapy and provides a predictive marker for determining if a patient suffering from metastatic melanoma or another cancer is likely to be a good responder to a treatment by Ipilimumab or another drug blocking an immune checkpoint.

Description

USE OF IMMUNE DIVERSITY AS A PREDICTIVE MARKER FOR IDENTIFYING PATIENTS LIKELY TO RESPOND TO AN ANTI-CTLA4 TREATMENT

Melanoma is a major public health problem worldwide, with an incidence doubling every 10 years. This disease, easily curable in early stages, has a bad prognosis once become metastatic. Response rate for metastatic melanoma with dacarbazine, which used to be the standard chemotherapy agent, is around 10%. Polychemotherapy approaches increase response rate, with no survival benefit. Korn et al. have reported in a meta-analysis of 42 phase II trials of metastatic melanoma, a median survival of 6.2 months (5.9 - 6.5 months) (Korn et al., 2008). In a multivariate analysis, some factors such as performance status, visceral metastasis, gender and brain metastasis, seem to impact the survival.

After decades of disappointing results, the recent successes obtained in clinical oncology by immunotherapies confirm the relevance of immune system stimulation in this setting. Antigen 4 of cytotoxic T lymphocytes (CTLA-4) is a negative regulator of T cell activation. An anti CTLA-4 antibody, also called Ipilimumab and marketed as Yervoy™, potentiates immune response leading to activation, proliferation of T lymphocytes and their invasion of tumour. This treatment allows a more efficient adaptive immune response against tumours. This indirect anti-cancer strategy, based on long-term immune system activation to fight the tumour, allowed for the first time to increase the life expectancy of patients with metastatic melanoma and to decrease the risk of relapse (Hodi et al., 2010).

Lymphopenia following treatment has been associated with response to Yervoy™ (Ku et al., 2010). Patients with an absolute lymphocyte count (ALC) >1000/μΕ after 2 Yervoy™ treatments (Week 7) had a significantly improved clinical benefit rate (51% vs. 0%; P = .01) and median OS (1 1.9 vs 1.4 months; P < .001) compared with those with an ALC <1000/nL.

Recent and promising clinical trials with an anti PD-1 and an anti-PD-Ll molecules in metastatic melanoma, non-small-cell lung cancer, colorectal cancer, renal-cell cancer, ovarian cancer, pancreatic cancer, gastric cancer, and breast cancer also showed that suppressing immune checkpoints could induce immune cells activation, thus promoting the development of an effective and durable anti-tumour adaptive response (Brahmer et al., 2012; Hamid et al., 2013; Ribas, 2012).

Unfortunately, all clinical studies on anti-CTLA4 treatment showed antitumor efficacy in only ~20% of patients, whereas they induce serious side effects. In addition, toxicity and efficacy were not correlated in a given patient. Given the high level of deleterious impact and the cost of such treatment, one of the main issues to be addressed by health care professionals is to understand why these new treatments work only for a small subset of patients. Thus, the analysis of involved immune mechanisms and the validation of predictive biomarkers represent a real challenges to better determine, before treatment, patients likely to respond to immunotherapy. Non-responder patients could then avoid heavy side effects and be directed earlier to other therapies. In addition, this could save unnecessary costs.

In this context, a number of scientists actively searched for a predictive marker of a patient's response to an immune checkpoint blockade therapy, and confusing results were recently published. In an international patent application filed in 2013 (WO 2014/055561), Harlan Robins et al. suggested that high TCR repertoire clonality at baseline (before the beginning of a treatment with an immune checkpoint blocker) could be a predictor of immunotherapy non-responder status. The results supporting this suggestion, however, were insufficient to identify any serious tendency, since they were obtained from only three patients, including two supposed responders. The group of three patients observed to derive this hypothesis was thus not representative of the patients usually treated by Ipilimumab. Noticeably, these results were already present in the priority application of WO 2014/055561, which was filed in 2012, and were not completed upon the filing of the international application, suggesting that these results could not be confirmed on a larger group of patients.

In 2014, a scientific publication (Robert et al., Clinical Cancer research 2014), signed, inter alia, by two of the inventors of WO 2014/055561, reported a study on 21 patients receiving an anti-CTLA4 monoclonal antibody on the immunomodulatory effects of CTLA4 blockade. In this publication, the authors clearly indicated that they did not observe any association between baseline TCR repertoire diversity and clinical response to CTLA4 blocking antibodies.

Hence, there is an important need for a predictive marker of a patient's response to an immune checkpoint blockade therapy, so that only the patients who are likely to really benefit from this therapy receive these potentially dangerous and expensive drugs.

As shown in the experimental part below, the inventors have now demonstrated that the diversity of the immune repertoire of TCR and BCR is a key parameter for predicting the response of a patient suffering from metastatic melanoma to an anti-CTLA4 antibody. Indeed, they observed that patients having a poor diversity of the immune repertoire at baseline fail to respond to the treatment, while those with a good diversity are most likely to respond to the treatment.

Hence, the present invention first pertains to an in vitro method of predicting the response of a cancer patient to treatment with a medicament or drug blocking an immune checkpoint, comprising: (i) measuring the diversity of the immune repertoire in a biological sample from said patient, and (ii) comparing the measured diversity to a predetermined threshold. A measured diversity lower than the threshold is indicative for a cancer patient who would not respond to the treatment. Of course, as for most predictive markers and despite the good specificity and sensitivity of the marker disclosed herein, "indicative for a patient who would not respond to the treatment" does not mean that this patient would not respond to the treatment with a 100% certainty. Rather, patients whose diversity is lower than the predetermined threshold are those who would most likely not respond to the treatment, i.e., the response rate of this group of patients is lower than that of the group of patients having a diversity higher than the threshold. For these patients, the risk- benefit balance is such that it is better not to treat them with a drug blocking an immune checkpoint, to avoid unnecessary side-effects and costs. Then the patient can be orientated toward a new molecule tested in a clinical trial, or a targeted therapy, or chemotherapy.

When performing the present invention, the "diversity of the immune repertoire" can be measured by collecting data comprising, for each rearrangement or clonotype observed in a sample, its relative abundance. The diversity is advantageously measured by determining the relative abundances of substantially all of the repertoire's rearrangements or clonotypes in an appropriate sample of T cells and/or B cells (such as a peripheral blood sample containing such cells).

As used herein, a "clonotype" means a recombined nucleotide sequence of a T cell or B cell encoding a T cell receptor (TCR) or B cell receptor (BCR), or a portion thereof, and a "rearrangement" encompasses all the recombined nucleotide sequences encoding a TCR or BCR or a portion thereof which comprise a given V segment and/or a given J segment. For example, the Vx rearrangement encompasses all the recombined nucleotide sequences encoding a TCR or BCR or a portion thereof which comprise the Vx segment, and the Vx-Jy rearrangement encompasses all the clonotypes which comprise the Vx and Jy segments, without considering the possible junctional differences between them. By extension, the terms "clonotype" and "rearrangement" may also designate, in the present text, the proteins encoded by such recombined nucleotide sequences (i.e., a V-(D)-J amino acid sequence or portion thereof, with a specific full amino acid sequence (clonotype) or with possible junctional mutations, deletions or additions (rearrangement)). Hence, a Vx rearrangement is the sum of all the Vx-Ji rearrangements (i taking all the possible values), and a Vx-Jy rearrangement is the sum of all the clonotypes having Vx and Jy. The clonotypes encompassed in a given Vx-Jy rearrangement can differ by their CDR3 region. The method according to the invention can hence be performed with different levels of repertoire profiles, the roughest being that obtained with V rearrangements only, and the most detailed or refined being the clonotype profile and/or at the sequence level. However, whatever the repertoire profile which is used, the principles and conclusions of the method are identical.

When performing the above method, patients whose measured diversity of the immune repertoire is higher than the predetermined threshold have a higher rate of response to the treatment, resulting in enrichment of responders in that group of patients.

The above method can be performed from any biological sample enabling the determination of the diversity of the immune repertoire. Non-limitative examples of such samples include a whole blood sample, a blood clot, PBMCs, a tissue biopsy, etc. According to a preferred embodiment of the present invention, TCR diversity is measured. In particular, the measured TCR diversity can include or consist of TCR combinatorial diversity. According to a particular way of conducting the method according to the invention, illustrated in the experimental part below, the diversity of the immune repertoire is measured by a multiplex PCR assay allowing the simultaneous detection of a significant number of TRBV-TRBJ rearrangements. For example, this assay can be designed so as to detect the presence or absence (and/or the level) of at least 20, preferably at least 35, at least 50, at least 100 or at least 200 TRBV-TRBJ rearrangements. Of note, the results shown in the experimental part below have been obtained by analyzing 276 different TRBV-TRBJ rearrangements covering 100% of the possible combinatorial rearrangements, but the skilled artisan can select, among these, the most informative ones, and/or use primers allowing the amplification of rearrangement representative of mono- or multi-gene families, in order to routinely perform the above-described method with a decreased complexity, for example by amplifying fragments from no more than 20 TRBV-TRBJ rearrangements.

It is important to note that in the frame of the present invention, several parameters of the immune diversity can be measured. In particular, the diversity can be expressed as a percentage of the number of rearrangements or clonotypes which can theoretically be observed by the technology used to measure it (called "global diversity" or "diversity richness" in the following text). For example, when the measured diversity is the combinatorial diversity and is measured by a PCR assay allowing the detection of 150 TRBV-TRBJ rearrangements if they are present in the sample, the global diversity of a sample in which 75 rearrangements are observed is 50%.

Another parameter of the immune diversity can advantageously be measured when performing the method according to the present invention is the percentage of observed rearrangements/clonotypes which are responsible for a certain amount (X%) of contribution amongst the observed rearrangements. This parameter, called "divX" or "EvennessX", reflects the evenness of the rearrangements/clonotypes present in the sample (which can also be called "homogeneity", "uniformity" or "consistency"). For example, div50 corresponds to the percentage of rearrangements/clonotypes necessary to reach 50% of contribution amongst observed rearrangements/clonotypes. DivX (wherein X is comprised between 0 and 100, preferably between 20 and 80) is a parameter of diversity allowing to characterize the level of evenness of an immune repertoire by focusing on the most important rearrangements/clonotypes in terms of contribution to the whole repertoire. DivX is the percentage of rearrangements necessary to reach X% of contribution amongst present rearrangements/clonotypes. The lower the value of divX, the less rearrangements/clonotypes contribute to X% of the repertoire diversity. Calculation of divX:

Rearrangements listed in order of contribution (the more abundant appears first in the list)

N first (i.e., N more important in terms of contribution) rearrangements are considered so that N is the smallest integer for which:

∑ N first rearrangements≥ X% contribution

EvennessX = Div X = N / number of rearrangements detected

In the above formula, the "rearrangements/clonotypes detected are those which are taken into account for calculating the richness of the repertoire, whatever their contribution to the repertoire.

Figure 1 illustrates div50 in two different samples.

According to a particular embodiment of the present invention, the diversity of the immune repertoire is expressed as the number of rearrangements necessary to reach X% of contribution amongst the observed rearrangements (DivX, also called EvennessX), with 20<X<80. For example, Div50 is measured. Of course, the skilled artisan will adapt the predetermined threshold to the parameter which is measured by assessing, in a given cohort, which couple (X, threshold) provides the best sensitivity and specificity (ROC curves can be used to this aim).

More generally, the threshold to be considered when performing the above method is predetermined by measuring the diversity of the immune repertoire in a representative cohort of individuals treated by an immune checkpoint blockade therapy, and whose response to this treatment is known. The threshold is calculated to obtain the best predictability (sensitivity and specificity) for the response. For example, as disclosed in the experimental part below, when the combinatorial diversity of the immune repertoire at baseline (i.e., before the treatment) is measured in a blood clot with a technology similar to that described in the experimental part and expressed as a percentage of observed rearrangements, a threshold of 65 to 95%, preferably 80 to 90% and for example around 85%, gives good results for predicting the response of patients suffering from metastatic melanoma to an anti-CTLA-4 antibody. In the same conditions, but if the diversity is expressed as div50, a predetermined threshold of 10 to 35%, for example around 25%, can be considered. In the same conditions, but if the diversity is expressed as div25, a predetermined threshold of 5 to 15%, for example 9.5%, can be considered. Of course, the skilled artisan is free to re-evaluate these thresholds on a larger cohort of patients, and by using any kind of technology for measuring the diversity of the immune repertoire (such as, for example, Next Generation Sequencing (NGS), spectratyping, immunoscope, Flow cytometry analysis...). The skilled artisan can also refine the threshold for particular subpopulations, depending on the type of cancer, the specific drug considered for blocking an immune checkpoint, or any other parameter regarding the patients, their pathologies (age of patient, associated treatment, type of comorbidities, ...) or their treatment (co-administration of another antineoplastic drug, different dosage regimen, etc.). Of course, the threshold also depends upon the parameter(s) used to assess the diversity of the immune repertoire (global diversity, Div25, Div50, etc.) and the skilled artisan can perfectly determine an optimal couple (parameter, threshold) or an optimal triplet (technology, parameter, threshold) to obtain a predictive method as sensitive and specific as possible.

Several new immunotherapies have been and are being developed to target the immune checkpoints. Ipilimumab (anti-CTLA4) was the first of these new therapeutics to be approved by the FDA in March 2011 for advanced melanoma, and other immunomodulators trials are ongoing in other cancers with similar encouraging results like with the anti PD-l/PD-Ll . Many other immune checkpoints have been identified and could potentially be targeted. Accordingly, the present method is useful for predicting the response of a cancer patient to treatment with any medicament targeting an immune checkpoint, and in particular with a drug chosen amongst anti-CTLA-4 antibodies, anti-PD-1 antibodies and anti- PD-L1 antibodies. More particularly, it can be used for predicting the response to an anti- CTLA-4 monoclonal antibody, especially an anti-CTLA-4 human IgGl such as Ipilimumab, or an anti-CTLA-4 human IgG2 such as Tremelimumab.

The immune system plays a dual role against cancer: it prevents tumor cell outgrowth and also sculpts the immunogenicity of the tumor cells. Drugs blocking an immune checkpoint can hence be used to treat virtually any type of cancer. Thus, the predictive method according to the invention is potentially useful for predicting a patient's response to such a drug of patients having a cancer selected amongst adrenal cortical cancer, anal cancer, bile duct cancer (e.g. periphilar cancer, distal bile duct cancer, intrahepatic bile duct cancer), bladder cancer, bone cancers (e.g. osteoblastoma, osteochondroma, hemangioma, chondromyxoid fibroma, osteosarcoma, chondrosarcoma, fibrosarcoma, malignant fibrous histiocytoma, giant cell tumor of the bone, chordoma, lymphoma, multiple myeloma), brain and central nervous system cancers (e.g. meningioma, astocytoma, oligodendrogliomas, ependymoma, gliomas, medulloblastoma, ganglioglioma, Schwannoma, germinoma, craniopharyngioma), breast cancer (e.g. ductal carcinoma in situ, infiltrating ductal carcinoma, infiltrating lobular carcinoma, lobular carcinoma in situ, gynecomastia), Castleman disease (e.g. giant lymph node hyperplasia, angiofollicular lymph node hyperplasia), cervical cancer, colorectal cancer, endometrial cancers (e.g. endometrial adenocarcinoma, adenocanthoma, papillary serous adnocarcinoma, clear cell), esophagus cancer, gallbladder cancer (mucinous adenocarcinoma, small cell carcinoma), gastrointestinal carcinoid tumors (e.g. choriocarcinoma, chorioadenoma destruens), Hodgkin's disease, non- Hodgkin's lymphoma, Kaposi's sarcoma, kidney cancer (e.g. renal cell cancer), laryngeal and hypopharyngeal cancer, liver cancers (e.g. hemangioma, hepatic-adenoma, focal nodular hyperplasia, hepatocellular carcinoma), lung cancers (e.g. small cell lung cancer, non-small cell lung cancer), mesothelioma, plasmacytoma, nasal cavity and paranasal sinus cancer (e.g. esthesioneuroblastoma, midline granuloma), nasopharyngeal cancer, neuroblastoma, oral cavity and oropharyngeal cancer, ovarian cancer, pancreatic cancer, penile cancer, pituitary cancer, prostate cancer, retinoblastoma, rhabdomyosarcoma (e.g. embryonal rhabdomyosarcoma, alveolar rhabdomyosarcoma, pleomorphic rhabdomyosarcoma), salivary gland cancer, skin cancer (e.g. melanoma, nonmelanoma skin cancer), stomach cancer, testicular cancers (e.g. seminoma, nonseminoma germ cell cancer), thymus cancer, thyroid cancers (e.g. follicular carcinoma, anaplastic carcinoma, poorly differentiated carcinoma, medullary thyroid carcinoma, thyroid lymphoma), vaginal cancer, vulvar cancer, and uterine cancer (e.g. uterine leiomyosarcoma). More particularly, the method according to the invention can be used for predicting a patient's response to a medicament targeting an immune checkpoint, wherein the patient has a cancer selected from the group consisting of metastatic melanoma, non-small cells lung carcinoma (NSCLC), small cell lung cancer (SCLC), prostate cancer and prostatic neoplasms (especially metastatic homrmone-refractory prostate cancer), sarcoma, Wilm's tumor, lymphoma, neuroblastoma, and bladder cancer.

As recalled above, several parameters of the diversity of the immune repertoire can be measured to characterize this repertoire. In particular, the global percentage of the diversity reflects its richness, whereas DivX parameters, such as Div25 or Div50, reflect its evenness. According to the present invention, these two parameters can be combined to more precisely predict a patient's response. When combining two parameters like this, a patient having both measured parameters above the corresponding predetermined thresholds will be most likely a good responder to the treatment, while a patient having only one parameter above the corresponding threshold will have an intermediate probability to respond to the treatment, and a patient having both parameters below the corresponding thresholds would most likely not respond to the treatment (and hence, will normally not receive it).

According to another embodiment, the method according to the present invention is carried out by combining the measure of the diversity of the immune repertoire to the absolute lymphocyte count (ALC). In this case, an ALC superior to a predetermined threshold (typically, ΙΟΟΟ/μί), in addition to a measured diversity above the corresponding predetermined threshold, is a further indicator that the cancer patient will most likely respond to the treatment.

The inventors have also demonstrated that DivX, in particular Div25, can be a prognostic marker for patients suffering from metastatic melanoma. According to another aspect, the present invention pertains to a method for in vitro determining if a patient suffering from metastatic melanoma has an increased risk of death, wherein a patient suffering from metastatic melanoma and having a Div25 < 9.6% at baseline has an increased risk of death and a patient suffering from metastatic melanoma and having a Div25 > 9.6% at baseline is likely to survive longer.

According to a further aspect, the present invention pertains to a method of treating a cancer patient, comprising a first step of performing a method as described above to predict the patient's response to an immune checkpoint blockade therapy, and in case the patients is identified as a good responder to such a therapy, a second step of administering a drug blocking an immune checkpoint to the patient.

The present application also pertains to a method of treating a cancer patient identified as a good responder to an immune checkpoint blockade therapy by a method as described above, comprising administering a drug blocking an immune checkpoint to the patient.

In the above treatment methods, the drug can be selected from any immune check point blockade drug, especially the group consisting of an anti-CTLA-4 antibody, and anti-PD-1 antibody and an anti-PD-Ll antibody. In particular, it can be an anti-CTLA-4 monoclonal antibody, especially an anti-CTLA-4 monoclonal IgGl such as Ipilimumab. According to a particular embodiment, the method is carried out by administering Ipilimumab to a responder patient at the dose of 3mg/kg every 3 weeks, for a total of 4 cures.

Of course, the same considerations as those mentioned above concerning the predicting methods also apply to the treatment methods. In particular, the patient can suffer from virtually any cancer. According to a particular aspect, the patient suffers from metastatic melanoma.

The present invention also pertains to a method of treating a cancer patient identified as a non responder to an immune checkpoint blockade therapy by a predictive method as described above, comprising administering dacarbazine to the patient.

The present invention also pertains to Ipilimumab, for use as a medicament for treating a patient suffering from metastatic melanoma and having a Div50 > 25% at baseline, and/or having a Div25 > 9.5% at baseline, and/or whose immune repertoire diversity is at least 85% at baseline.

As described in Example 9 below, the inventors observed that patients who responded to a treatment with Ipilimumab exhibited a profile of V(D)J rearrangements statistically distinct from the non-responders. Hence, another aspect of the present invention is an in vitro method of predicting the response of a cancer patient to a treatment with a drug blocking an immune checkpoint, comprising: measuring a diversity of the immune repertoire in a biological sample from said patient, calculating the contribution of one or several observed rearrangement(s) in said diversity, thereby generating a profile of the patient's immune repertoire and calculating the distance between the obtained profile and one or several predetermined profiles. For performing this method, the predetermined profile(s) can comprise at least one profile representative of non-responder patients and/or at least one profile representative of responder patients and/or at least one profile representative of healthy individuals. These profiles can also be representative of only a category of responders or non- responders, defined, in addition to their response to a treatment, by parameters such as the kind and stage of their cancer, physiological parameters from the patients (sex, age, etc.) or parameters resulting from the medical and treatment history of the patients (surgery, previous therapy, previous observed resistance to a treatment, etc.).

According to a particular embodiment of the above method, the predetermined profile(s) comprise at least one profile representative of (at least a category of) non-responder patients, and a small distance between the patient's immune repertoire profile and this predetermined profile (for example, below a predetermined threshold) indicates that this patient will probably not respond to the treatment.

According to another particular embodiment of the above method, the predetermined profile(s) comprise at least one profile representative of (at least a category of) responder patients, and a small distance between the patient's immune repertoire profile and this predetermined profile (for example, below a predetermined threshold) indicates that this patient will probably respond to the treatment.

According to yet another particular embodiment of the above method, the predetermined profile(s) comprise at least one profile representative of healthy individuals and a small distance between the patient's immune repertoire profile and this predetermined profile (for example, below a predetermined threshold) indicates that this patient will probably respond to the treatment.

According to yet another particular embodiment of the above method, the predetermined profiles comprise one profile representative of (at least a category of) non- responder patients and one profile representative of (at least a category of) responder patients, and the patient will be considered as a non-responder if the distance between his/her profile and that of (his/her category of) non-responders is smaller than the distance between his/her profile and that of (his/her category of) responders.

According to a particular embodiment of the method, the profiles consist of the contribution of only one V(D)J rearrangement to the immune repertoire diversity. In such a case, the distance between the measured profile and the predetermined profile(s) is hence, naturally, the difference between the values of the contribution of this V(D)J rearrangement to the immune repertoire diversity. When the method is performed with profiles comprising several distinct V(D)J rearrangements, several methods can be used to calculate the distance between two profiles. For example, a weight can be attributed to each of the rearrangements of the profiles, in order to determine a weighted barycentre of the profile, and the distance between the profile is then the distance between the two barycentres. Alternatively distance can be calculated by correlation method, Euclidean method, maximum method, Manhattan method, Canberra method, binary method, Minkowski method ...

In the embodiments of the above method where the distance between profiles is compared to a predetermined threshold, this threshold will be determined by the skilled artisan by assessing, in a given cohort and with a determined combination of rearrangements included in the profiles, which threshold provides the best sensitivity and specificity (ROC curves can be used to this aim).

As described in Example 9 below, the inventors have identified 5 V-J rearrangements which are statistically more frequent in responders to Ipilimumab than in non- responders, and 3 V-J which are statistically less frequent in non-responders than in responders. A method as above described can hence advantageously be performed by including, in the immune repertoire profiles, the contribution of the following rearrangements in the measured combinatorial diversity: V10-J1.4, V15-J1.6, V13-J1.4, V2-J2.4, V11-J2.3,

V16-J2.4, V16-J2.1 and V14-J2.4. According to this embodiment, in a predetermined profile representative of responders, the rearrangements V10-J1.4, V15-J1.6, V13-J1.4, V2-J2.4 and

V11-J2.3 are preferably more frequent and the rearrangements V16-J2.4, V16-J2.1 and V14-

J2.4 are preferably less frequent, compared to a profile representative of non-responder patients.

Of course, the skilled artisan is free to re-evaluate the relevancy of these and other rearrangements on a larger cohort of patients, and by using any kind of technology for measuring the diversity of the immune repertoire (such as, for example, Next Generation Sequencing (NGS), spectratyping, immunoscope, Flow cytometry analysis...). The skilled artisan can also refine the profiles (nature of the observed rearrangements and values of their contributions to the diversity) for particular subpopulations, depending on the type of cancer, the specific drug considered for blocking an immune checkpoint, or any other parameter regarding the patients, their pathologies (age of patient, associated treatment, type of comorbidities, ...) or their treatment (co-administration of another antineoplastic drug, different dosage regimen, etc.).

In Example 10, the inventors observed a drop of diversity after ipilimumab treatment in a patient initially considered as likely to be a good responder (diversity richness above the threshold calculated in the considered cohort), but who eventually failed to respond to the treatment. They deduced that kinetic analysis of the immune repertoire diversity can be used as a predictive marker of response to treatment with a drug blocking an immune checkpoint.

The present invention hence also pertains to an in vitro method of evaluating the response of a cancer patient to treatment with a drug blocking an immune checkpoint, comprising: measuring a diversity of the immune repertoire in at least two biological samples from said patient, obtained at two different time points, wherein at least one time point is after administration of at least one dose of said treatment, wherein a decrease in said diversity indicates that the patient does not respond to the treatment. Optionally, the evolution of the diversity between the two time points can be calculated and compared to a predetermined threshold. The patient is then considered as not responding to the treatment if a decrease above this threshold is observed between the two time points.

This method is particularly useful for confirming or infirming a prediction of the patient's response, based on a first measure of diversity which led to the conclusion that the patient could be a good responder. Indeed, if the diversity measured in the first biological sample is above a predetermined threshold, but if this diversity then decreases, this indicates that the patient does not respond to the treatment and must be reclassified.

According to a preferred embodiment of the above method, the first time point is 0 to 5 days before or after the administration of the first dose of the treatment, and the second time point is 0 to 5 days before or after the administration of the third dose of said treatment. In particular, the diversities can be measured the same day as the administration of the doses, which limits the number of times the patient needs to come to the hospital.

Detailed characteristics have been mentioned above as specific embodiments of the method of predicting the response of a cancer patient to treatment with a drug blocking an immune checkpoint by comparing the measured diversity to a predetermined threshold, such as the nature of the biological sample used to perform the method, the technology used to measure the diversity, the combination of the absolute diversity and the evenness, the combination of the diversity and the absolute lymphocyte count etc. Of course, all these characteristics can also apply to the other methods described in the present text.

The present invention will be understood more clearly from the further description which follows, which refers to examples illustrating the predictive value of combinatorial diversity to determine, at baseline, if a patient having a metastatic melanoma is likely to respond to a treatment by an anti-CTLA4 antibody, as well as to the appended figures.

Figure 1 : In example A), 36 amongst 235 pics allow to reach 50% of the contribution (15,3% of the 235 pics). In example B), 70 amongst 255 pics allow to reach 50% of the contribution (27,4% of the 255 pics).

Figure 2: Comparison of TRB diversity levels at DO between responders (R) and non responders (NR.) patients (Wilcoxon rank sum test, p-value=0,033).

Figure 3: ROC analysis for response assessment as a function of combinatorial diversity.

Figure 4: Patient's response according to combinatorial diversity. Pearson's

Chi-squared test with Yates' continuity correction (p-value= 0.147)

Figure 5: Comparison of div50 levels at DO between responders (R) and non responders (NR) patients (Wilcoxon rank sum test, p-value=0,028). Figure 6: ROC analysis for response assessment as a function of div50.

Figure 7: Div50 as a function of hTRB diversity.

Figure 8: Comparison of div25 levels at DO between responders (R) and non responders (NR) patients (Wilcoxon rank sum test, p-value=0.028)

Figure 9: ROC analysis for response assessment as a function of div25

Figure 10: Patient's response according to div25. Pearson's Chi-squared test with Yates' continuity correction (p-value=0.066)

Figure 1 1 : Analysis of Overall Survival (OS) according to baseline div25

(<9.6% or >9.6%) (log rank p-value = 0.0478).

Figure 12: Patients' early death according to div25 (<9.6% or >9.6%).

Pearson's Chi-squared test with Yates' continuity correction (p-value = 0.488).

Figure 13: Div50 a function of global combinatorial TCR diversity.

Figure 14: Absolute Lymphocyte Count (ALC) as a function of div50.

Figure 15: Dendrogram of 12 samples at Dose 1 generated by hierarchical clustering (NR = non-responders; R = responders). The data are the expression of 276 rearrangements. The measure of distance used for clustering was based on correlation. Ward method was employed as the method of agglomeration.

Figure 16: Volcano-plot representation of median contribution difference between responders (R) and non-responders (NR) to treatment in paired samples and corresponding p-value, for each of the 276 TRB VJ rearrangements; a dot represents the difference: NR - R treatment contribution of one specific TRB VJ rearrangement and its p- value. P-values were assessed using Wilcoxon rank sum test. Of note, despite multiple comparisons, Bonferroni correction was not applied due to weak statistical power.

Figure 17: Volcano-plot representation of median contribution evolution between dose 1 (Dl) and dose 3 (D3) treatments in paired samples and corresponding p- value, for each of the 276 TRB VJ rearrangements; a dot represents the difference: D3 - Dl treatment contribution and its p-value of one specific TRB VJ rearrangement. P-values were assessed using Wilcoxon signed-rank test and rearrangements with significant differences (p- value < 0.05, above the red line) between groups are identified. Of note, despite multiple comparisons, Bonferroni correction was not applied due to weak statistical power.

Figure 18: re stratification of good diversity patients (i.e., global diversity >85%) based on diversity evolution between Dose 1 and Dose 3: One patient out of 7 had a drastic decrease of diversity (-18.1%) between Dose 1 and Dose 3, which could explain the absence of response. This patient could then be re stratified in non-responder group.

EXAMPLES

1. Materials and Methods

Response assessment Response was assessed by practitioner according to their expertise. Patient's selection

All the following results have been obtained on the same cohort. Patients were classified into a response group according to physician expertise. Responders correspond to patients who showed particularly high clinical benefit. Non-responders correspond to patients who showed no clinical benefit at all.

Cohort characteristics:

No difference could be observed in div50<25% and div50 >25% arms in terms of age, LDH rate and absolute lymphocyte count (ALC) (Table 1). All the patients of that study had a Karnofsky index >90%.

N Min Median Mean Max Sd C.V. Wilcoxon test p-value

Age (year) Div50 <25% 7 52 68 67,9 77 8,45 0,125 0,81

Div50 >25% 5 33 66 58,2 78 21,29 0,366

LDH Div50 <25% 6 183 211 245,2 438 96,45 0,393 0,29

Div50 >25% 1 156 156 156 156

ALC Div50 <25% 7 0,4 1 1 1,5 0,37 0,372 0,06

Div50 >25% 5 1,1 1,5 1,4 1,9 0,34 0,239 Combinatorial diversity analysis (Multiplex PCR assay).

Genomic DNA was extracted from blood clot using QIAamp DNA Blood Mini Kit (Qiagen) and concentrated using Amicon ultra 0.5 mL Centrifugal Filters (Millipore). Multiplex PCR was performed using an upstream primer specific of all functional members of a given V family and a downstream primer specific of a given J segment, as described in the patent application WO2009/095567. This assay allows the simultaneous detection of 276 different TRBV-TRBJ rearrangements covering 100% of the possible combinatorial rearrangements. All V-Jl, J2, J3, J4, Jn products were separated as a function of their size with a maximum amplicon size expected of ~5 kb. PCR signals were detected and analyzed using the Constel'ID software developed by ImmunID Technologies (Grenoble, France). It is here important to mention that the multiplex PCR strategy could potentially amplify a DNA fragment corresponding to a rearrangement involving a Vx+1 gene located 3' of the targeted Vx gene. However, these side products were recognized and excluded from the final results by the Constel'ID software.

PCR products were generated using iProof GC rich Master Mix (Bio-Rad) with cycling condition as follows: 98°C for 3 min, 98°C for 20 sec, 72°C for 20 sec and 72°C for 3 min 30 sec and reducing the annealing temperature by one degree every cycle until 68°C; last cycle at 68°C; final temperature was then repeated 27 times. Finally, one cycle of 3 min at 72°C was performed. In order to normalize DNA quantity in each reaction, the actin gene was amplified in the same PCR run. PCR products were purified by ultrafiltration (Nucleofast 96, Macherey-Nalgen) and separated on Lab on chip (HT DNA 12K LabChip Kit), by microfluidic migration (Labchip GX, PERKIN ELMER). ConstePID software was used to analyse all the data provided by the migration.

Statistical analysis

Results are presented as box plots with individual values. Comparisons between groups were made using the nonparametric Mann- Whitney U test (response vs. non response). The nonparametric Wilcoxon's paired test was used to assess variations between time points. Fisher's exact test or Yates' Chi2 test was used to compare proportions. Spearman's correlation test was used to assess correlations between TCR diversity values and clinical and biological variables. A receiver operating characteristic curve (ROC) was performed to determine the cutoff values for TCR diversity and div50 with regard to prediction of response to Yervoy. The best cutoff value was selected based on likelihood ratio and Youden's index. Statistical analyses were performed using the free statistical software package R. A p value < 0.05 was considered statistically significant.

2. Predictive value of the combinatorial TCR diversity

In a recent publication (CTLA4 Blockade Broadens the Peripheral T-Cell Receptor Repertoire. Robert L Clin Cancer Res. 2014 Apr 9) the analysis of the immune repertoire at the CDR3 level (NGS) was not able to predict response to anti-CTLA4 patient.

In a case study, the inventors now compared combinatorial diversity of 8 non responders & 4 responder patients to Yervoy™ (BMS anti-CTLA-4) immunotherapy in metastatic melanoma. Analysis performed on blood clot showed that response was associated with better diversity measured at baseline (Figure 2)(Wilcoxon rank sum test, p-value=0,033). A ROC analysis was performed and 85% diversity was chosen as threshold in order to optimize specificity (Figure 3). Surprisingly, no patient having a combinatorial diversity <85% (n=5) responded to the Yervoy™. At the opposite, 4 patients out of 7 with a diversity >85% diversity responded to Yervoy™ (Figure 4).

3. Predictive value of div50

Combinatorial diversity was further assessed through div50 analysis and showed a positive correlation between response to treatment and evenness of the repertoire at baseline. Indeed, responder patients had a significantly higher div50 compared to non- responders (Figure 5)(Wilcoxon rank sum test, p-value=0,028). A ROC analysis was performed and 25% div50 was chosen as threshold (Figure 6). No patient having restricted repertoire, i.e., div50 <25% (n=7) responded to the Yervoy™. To the contrary, 4 patients out of 5 with a div50 >25% responded to Yervoy™ (Figure 7)(Pearson's Chi-squared test with Yates' continuity correction (p-value= 0.023). These results show that a diversified immune lymphocyte repertoire measured in peripheral blood is necessary in order to respond to Yervoy™ immunotherapy in metastatic melanoma.

4. Predictive value of div25

Combinatorial diversity was also assessed through div25 analysis and showed a positive correlation between response to treatment and evenness of the repertoire at baseline. Indeed, responder patients had a significantly higher div25 compared to non- responders (Fig. 8)(Wilcoxon rank sum test, p-value=0,028). A ROC analysis was performed and 9.5% div25 was chosen as threshold (Fig. 9). No patient having a div25 <9.5% (n=6) responded to the Yervoy™. To the contrary, 4 patients out of 6 with a div25 >9.5% responded to Yervoy™ (Fig. 10)(Pearson's Chi-squared test with Yates' continuity correction (p-value= 0.023)).

These results show that a diversified immune lymphocyte repertoire measured in peripheral blood is necessary in order to respond to Yervoy™ immunotherapy in metastatic melanoma.

5. Pronostic value of div25

An overall survival analysis show a positive correlation between a div25 >9.6% and a better survival (Fig. 11) (log-rank p- value = 0.0478). Only 1 patient out of 5 with a good div25 (i.e. >9.6%) deceased before one year. To the contrary, 4 patients out of 7 with a low div25 (i.e. <9.6%) deceased before 1 year (Fig. 12).

These results pinpoint the importance of diversity analysis as a biomarker to stratify patients according to their life expectancy.

6. Predictive value of the combination of Div50 and global combinatorial TCR diversity

The study of Div50 as a function of diversity highlights the potential of that approach for a better stratification of responders and non-responders to Yervoy™ immunotherapy. Responders seem to have both a good div50 and diversity, as illustrated in Fig. 13. Even if Div50 stratification alone is enough in order to stratify responders in the cohort studied herein, the combination of diversity and div50 could help by giving a more accurate picture of the immune status of the patient.

7. Predictive value of the combination of Div50 and global Lymphocyte count (ALC)

The study of Div50 as a function of Absolute Lymphocyte Count (ALC) highlights the potential of that approach for a better stratification of responders and non- responders to Yervoy™ immunotherapy. Responders seem to have both a good div50 and ALC as illustrated in Fig. 14. Even if Div50 stratification alone is enough in order to stratify responders in the cohort studied herein, the combination of ALC and div50 could help by giving a more accurate picture of the immune status of the patient.

8. Hierarchical clustering

Still using the results of the cohort described in paragraph 1 above, the inventors assessed whether one or several rearrangement(s) over-represented are necessary for a proper efficiency of immunotherapies such as Ipilimumab through specific recognition of tumor antigens. Such a signature of rearrangements could be used in order to stratify responders and non-responders patients.

To this aim, similarity between samples was measured based on correlation using each rearrangement contribution. Ward method was employed as the method of agglomeration. Unsupervised hierarchical clustering with multiscale bootstrap resampling of the samples was then performed in order to group samples based on their similarity.

As illustrated in Fig. 15, hierarchical clustering allowed the identification of a well distinct branch composed solely of non-responder patients.

9. Volcano plots

Volcano plot allows the identification of any single rearrangement significantly over-expressed or under-expressed compared to a referential.

Such rearrangements could be used (independently or combined with others rearrangements) as predictive markers of response to immunotherapies such as Ipilimumab. Biomarkers could be either a (group of) rearrangement(s) present at baseline, which would be over or under-represented when compared to a referential, or a (group of) rearrangement(s) significantly increasing or decreasing along treatment.

A first volcano plot was obtained with the median contribution difference between responders (R) and non-responders (NR) to treatment, for each of the 276 TRB VJ rearrangements. Rearrangements with significant differences between groups were identified (Fig. 16). In the cohort described in paragraph 1, the rearrangements over-represented in responders were: V10-J1.4, V15-J1.6, V13-J1.4, V2-J2.4 and V11-J2.3; the rearrangements under-represented in responders were: V16-J2.4, V16-J2.1 and V14-J2.4.

A second volcano plot was obtained with the median contribution evolution between paired samples, for each of the 276 TRB VJ rearrangements. Rearrangements with significant evolution between Dose 1 and Dose 3 of Ipilimumab were identified (Fig. 17). In the cohort described in paragraph 1, the contribution of V30-J2.6, V07-J1.3, V14-J1.6, V19- J2.6, V05-J2.6, V15-J1.5 and V10-J2.5 increased, while that of V30-J1.6, V14-J2.2, V28- J1.1/J1.2, V24-J1.1/J1.2, V15-J1.6, V24-J1.6, V06-J1.4, V02-J1.3 decreased. This evolution in repertoire contribution of rearrangements can also be monitored by correlation analysis (r²) in order to assess the level of modification of the repertoire between two time points or between a patient and a reference sample.

10. Kinetic analysis

Kinetic analysis has been performed between dose 1 and dose 3 of Ipilimumab treatment. A drop of diversity after ipilimumab treatment in non-responder patients could be due to i) a global degradation of the condition of the patient, and/or ii) the exaggerate proliferation of potentially self-specific clones leading to autoimmune disorders.

The evolution between 2 time points was assessed by difference of diversity richness between time point 2 and time point 1. A decrease of diversity was considered significant when two times superior to ImmunTraCkeR test technological variation (i.e., +/- 8%).

Although no firm conclusion can be drawn from the results due to the low number of patients, it is interesting to mention that a drastic decrease of diversity (superior to technological variation) between dose 1 and dose 3 in a patient from the good prognostic group (i.e., div >85%) could explain the absence of response for this patient (Fig. 18). Thus, this example highlights the potential of kinetic analysis as a predictive marker of response to ipilimumab.

REFERENCES

Brahmer, J.R., Tykodi, S.S., Chow, L.Q.M., Hwu, W.-J., Topalian, S.L., Hwu, P., Drake, C.G., Camacho, L.H., Kauh, J., Odunsi, K., et al. (2012). Safety and activity of anti-PD-Ll antibody in patients with advanced cancer. N. Engl. J. Med. 366, 2455-2465.

Galon, J., Costes, A., Sanchez-Cabo, F., Kirilovsky, A., Mlecnik, B., Lagorce-Pages, C, Tosolini, M., Camus, M., Berger, A., Wind, P., et al. (2006). Type, density, and location of immune cells within human colorectal tumors predict clinical outcome. Science 313, 1960-1964.

Hamid, O., Robert, C, Daud, A., Hodi, F.S., Hwu, W.-J., Kefford, R.,

Wolchok, J.D., Hersey, P., Joseph, R.W., Weber, J.S., et al. (2013). Safety and Tumor Responses with Lambrolizumab (Anti-PD-1) in Melanoma. N. Engl. J. Med.

Hodi, F.S., O'Day, S.J., McDermott, D.F., Weber, R.W., Sosman, J.A., Haanen, J.B., Gonzalez, R., Robert, C, Schadendorf, D., Hassel, J.C., et al. (2010). Improved Survival with Ipilimumab in Patients with Metastatic Melanoma. N. Engl. J. Med. 363, 711- 723.

Korn, E.L., Liu, P.-Y., Lee, S.J., Chapman, J.-A.W., Niedzwiecki, D., Suman, V.J., Moon, J., Sondak, V.K., Atkins, M.B., Eisenhauer, E.A., et al. (2008). Metaanalysis of phase II cooperative group trials in metastatic stage IV melanoma to determine progression-free and overall survival benchmarks for future phase II trials. J. Clin. Oncol. Off. J. Am. Soc. Clin. Oncol. 26, 527-534.

Manuel, M., Tredan, O., Bachelot, T., Clapisson, G., Courtier, A., Parmentier, G., Rabeony, T., Grives, A., Perez, S., Mouret, J.-F., et al. (2012). Lymphopenia combined with low TCR diversity (divpenia) predicts poor overall survival in metastatic breast cancer patients. Oncoimmunology 1, 432-440.

Pages, F., Berger, A., Camus, M., Sanchez-Cabo, F., Costes, A., Molidor, R., Mlecnik, B., Kirilovsky, A., Nilsson, M., Damotte, D., et al. (2005). Effector memory T cells, early metastasis, and survival in colorectal cancer. N. Engl. J. Med. 353, 2654-2666.

Ribas, A. (2012). Tumor Immunotherapy Directed at PD-1. N. Engl. J. Med. 366, 2517-2519.

Claims

1. An in vitro method of predicting the response of a cancer patient to treatment with a drug blocking an immune checkpoint, comprising: measuring diversity of the immune repertoire in a biological sample from said patient and comparing the measured diversity to a predetermined threshold, wherein a measured diversity lower than said threshold is indicative for a cancer patient who would not respond to the treatment, and a measured diversity higher than said threshold is indicative for a cancer patient who will respond to the treatment.

2. The method according claim 1, wherein the diversity of the immune repertoire is expressed as the number of rearrangements necessary to reach X% of contribution amongst the observed rearrangements (DivX), wherein X is comprised between 20 and 80.

3. The method according to claim 1 or claim 2, wherein a patient suffering from metastatic melanoma and having a combinatorial TCR diversity < 85% at baseline is most likely not to respond to Ipilimumab, and a patient suffering from metastatic melanoma and having a combinatorial TCR diversity > 85% at baseline is likely to respond to Ipilimumab.

4. The method according to any of the preceding claims, wherein a patient suffering from metastatic melanoma and having a Div50 < 25% at baseline is identified as a non-responder to Ipilimumab, and a patient suffering from metastatic melanoma and having a Div50 > 25% at baseline is identified as a responder to Ipilimumab.

5. The method according to any of the preceding claims, wherein a patient suffering from metastatic melanoma and having a Div25 < 9.5% at baseline is identified as a non-responder to Ipilimumab, and a patient suffering from metastatic melanoma and having a Div25 > 9.5%o at baseline is identified as a responder to Ipilimumab.

6. An in vitro method of predicting the response of a cancer patient to treatment with a drug blocking an immune checkpoint, comprising: measuring a diversity of the immune repertoire in a biological sample from said patient, calculating the contribution of one or several observed rearrangement(s) in said diversity, thereby generating a profile of the patient's immune repertoire and calculating the distance between the obtained profile and one or several predetermined profile(s).

7. The method according to claim 6, wherein the predetermined profile(s) comprise at least one profile representative of non-responder patients, and wherein if the distance between the patient's immune repertoire profile and this predetermined profile is below a predetermined threshold, the patient is considered as likely not to respond to the treatment.

8. The method according to claim 6 or claim 7, wherein the predetermined profile(s) comprise at least one profile representative of responder patients, and wherein if the distance between the patient's immune repertoire profile and this predetermined profile is below a predetermined threshold, the patient is considered as likely to respond to the treatment.

9. The method according to any of claims 6 to 8, wherein the patient's immune repertoire profile comprises the contribution of the following rearrangements in the measured combinatorial diversity: V10-J1.4, V15-J1.6, V13-J1.4, V2-J2.4, V11-J2.3, V16- J2.4, V16-J2.1 and V14-J2.4.

10. The method according to claim 9, wherein in a profile representative of responder patients, the rearrangements V10-J1.4, V15-J1.6, V13-J1.4, V2-J2.4 and V11-J2.3 are over-represented and the rearrangements V16-J2.4, V16-J2.1 and V14-J2.4 are under- represented, compared to a profile representative of non-responder patients.

1 1. The method according to any of the preceding claims, wherein said drug blocking an immune checkpoint is selected from the group consisting of an anti-CTLA-4 antibody, an anti-PD-1 antibody and an anti-PD-Ll antibody.

12. The method of claim 11, wherein said anti-CTLA-4 antibody is an anti-

CTLA-4 monoclonal antibody.

13. The method of claim 12, wherein said anti-CTLA-4 monoclonal IgGl is

Ipilimumab.

14. The method according to any of the preceding claims, wherein said cancer patient suffers from metastatic melanoma.

15. The method according to any of the preceding claims, wherein said biological sample is selected from the group consisting of a whole blood sample, a blood clot, PBMCs and a tissue biopsy.

16. The method according to any of the preceding claims, wherein said measured diversity is TCR diversity.

17. The method according to any of the preceding claims, wherein said measured TCR diversity is combinatorial TCR diversity.

18. The method according to any of the preceding claims, wherein said measured diversity of the immune repertoire is measured by a multiplex PCR assay allowing the simultaneous detection of at least 20 TRBV-TRBJ rearrangements.

19. The method according to any of the preceding claims, wherein the diversity of the immune repertoire is combined to the absolute lymphocyte count (ALC), and wherein an ALC superior to a predetermined threshold of

is indicative for a cancer patient who will respond to the treatment.

20. The method according to any of the preceding claims, wherein the diversity of the immune repertoire is expressed both as DivX with X comprised between 25 and 50 and as a global percentage of the diversity, and wherein the two results are combined.

21. An in vitro method of evaluating the response of a cancer patient to treatment with a drug blocking an immune checkpoint, comprising: measuring a diversity of the immune repertoire in at least two biological samples from said patient, obtained at two different time points, wherein at least one time point is after administration of at least one dose of said treatment, wherein a decrease in said diversity indicates that the patient does not respond to the treatment.

22. The method according to claim 21, wherein the diversity measured in the first biological sample is above a predetermined threshold.

23. The method according to claim 21 or claim 22, wherein the first time point is 0 to 5 days before or after the administration of the first dose of the treatment, and the second time point is 0 to 5 days before or after the administration of the third dose of said treatment.