FIELD OF THE INVENTION

The field of this invention is medical research in immune system boosting to fight cancer or infectious diseases.

More precisely, it relates to a method for rapidly, cheaply and reliably processing raw data generated from measurement of quantifiable biological parameters.

BACKGROUND OF THE INVENTION

The treatment of cancer is currently based on surgery, irradiation and antitumor drugs. Studied for decades, and now in experimental phase, the anti-cancer immunotherapy is a promising new approach which should reinforce the therapeutic arsenal against cancer.

Human organism is able to detect cancerous cells and other pathogen elements as foreign elements, and to send an immune response to destroy it. In most cases, tumour cells are destroyed by our lymphocytes before they evolve into a detectable cancer. However, if this response is insufficient, cancer could breakout.

The aim of anti-cancer immunotherapy is to stimulate the human immune system to reject and destroy tumors in a more sustained way than with conventional chemical drugs. Chemotherapy methods are associated to significant and sometimes severe adverse effects which can significantly alter patients' life. Moreover, a small number of undetectable cancerous cells sometimes survives, and can metastasis or continue to grow, resulting in cancer relapse after some months or years, whereas boosting or stimulating the patient's immune system by immunotherapy enhances the chance of obtaining durable cancer regression.

The therapeutic potential of immunotherapy is also promising for treating infectious diseases. In particular, several infectious diseases, and in particular viral diseases (HIV, HCV, etc ..) or parasitic disease (malaria for instance), involve a chronic phase due to the persistence of the infectious agent despite of chemical treatment. Boosting or stimulating the patient's immune system by immunotherapy is expected to break down such persistence.

Generally, immunotherapy aims at stimulating humoral and/or cellular immune responses towards one or more particular antigens expressed by cancer cells or infected cells.

At present time, many immunotherapy or gene therapy products are available in the art, with various compositions (e.g. peptide, protein antigens, viral or non-viral vectors expressing the antigen(s), cell composition which can be used independently or in combination, and with or without adjuvant, etc .), and a vast number are currently under test in preclinical or clinical studies. For example, the anti-tumoral efficacy of an anti-cancer candidate (TG4023) was demonstrated in different murine models, and will be tested in clinic in hepatocarcinoma patients. TG4023 is a recombinant vaccinia virus expressing nucleotide sequence encoding a yeast conditional suicide gene that efficiently converts the nontoxic compound 5-FC into 5-FU and 5-FUMP lethal metabolites. 5-FU can penetrate tumor cells, resulting in tumor cell killing while the patient's antitumor immune response can be enhanced through the local release of cellular debris that can then be processed and presented by the antigen- presenting cells.

More generally, an immunotherapy product can be administered for therapeutic vaccination, i.e. after cancer or infection has developed in the body, with the aim of inducing T cell-mediated responses, and particularly T cell-mediated cytotoxicity and cytokine responses, thus resulting in the elimination of cancer cells or infected cells. In addition, immunotherapy products can also be used as prophylactic vaccines for preventing infection or cancer occurrence. In this case, it is particularly useful to stimulate antibody responses.

A number of clinical studies are ongoing with various immunotherapeutic products targeting tumor-associated or viral antigens and encouraging data were recently reported which support the expected mechanism of action of such immunotherapy products. For example, the decrease of viral load and an increase of immune response were monitored in HCV-infected patients treated with a MVA (Modified Virus Ankara) vector expressing non structural HCV proteins NS3, NS4 and NS5B (described in WO2004/1 1 1082). On the same line, effective anti-tumor responses resulting in longer survival rate were reported in patients with non small cell lung cancer treated with a MVA vector expressing the MUC1 tumor- associated antigen together with IL-2 (TG 4010) in combination with chemotherapy (Velu et al., 2005, ASCO poster abstract; Raumlau et al., J Thorac Oncol. 2008;3(8):941 ). Encouraging data were also reported in the treatment of women with high grade cervical intraepithelial neoplasia (CIN2/3) injected with an immunotherapy product (TG4001/R3484) which targets the oncogenic E6 and E7 proteins of HPV (described in WO99/03885). One may also highlight the antitumor responses observed in patients with advanced cutaneous T-cell lymphoma (CTCL) and multilesional cutaneous B-cell lymphoma (CBCL) following intra-tumoral administrations of an adenovirus vector expressing the human interferon gamma (IFNg) gene (Urosevic et al, 2007, Curr. Opin Investig Drugs 8:493).

Tests for validating a particular candidate, either prophylactic or therapeutic, comprise a pre-clinical phase including research on animal models, and clinical trials with increasing numbers of patients tested for safety (phase I) and efficiency (phases II, III and IV) of the candidate vaccine. During each phase, not only the clinical response (improvement of the disease symptoms, reduction of the tumor size, viral clearance or reduction of the viral load, etc.) is monitored, but also many biological parameters. In particular, parameters which correlate with induction or stimulation of an effective specific immune response against the antigen(s) present in or expressed by the vaccine candidate are measured, in animals or patients treated and at several time points. Such antigens include the targeted antigens (e.g. expressed by the cancer or infected cells) and possibly some additional antigens such as those associated with the vaccine candidate or with the pathological model (e.g. MVA antigens when the vaccine candidate is based on a MVA vector). It can also be useful to measure non-specific immunity (e.g. cytokine or chemokine levels which correlate with activation of effector cells).

In the case of therapeutic vaccination, such biological parameters include the expression levels of a high number of molecules including but not restricted to cytokines and chemokines, inflammatory factors, signaling molecules or cytotoxic molecules such as perforin, granzyme or FasL. Cytokines and chemokines are proteins playing a role of messenger between cells of the immune system to coordinate the whole immune response. Examples of chemokines and cytokines that may be monitored for assessing the biological efficacy of a particular immunotherapy product include without limitation RANTES, MIP-1 a, MIP-1 b, Eotaxin, IP10, MCP-1 , IL-7, IL-10, IL1 -ra, EGF, IL-1 a, IL-1 b, TNFa, IL-13, IL-5, GM-CSF, IFNg, IL- 12, IL-4, IL-2, IL-8, IL-17, IL-15, and sCD40L.

In addition, the monitoring of the immune response may involve measures of the cytotoxic capacity of T CD8 or CD4 cells, which may include measures of the expression levels of cytotoxic molecules such as perforin, granzyme or FasL by effector cells, or direct measures of the amount of cytotoxicity obtained when effector cells are contacted with target cells expressing the antigen present in the immunotherapy product.

When the vaccine candidate is for prophylactic use (i.e. for preventing infection or cancer), the monitoring of the immune response may comprise the measure of the concentration of antibodies specific for antigens present in or expressed by the vaccine candidate in serum, as well as the evaluation when possible of the activity of the stimulated antibodies (e.g. neutralizing or not).

These immune response parameters may be measured using various technologies well known to those skilled in the art.

However, no matter which technology is used, in view of the number of animals and patients tested, the presence of several time points, the necessity to make the measures at several dilutions, and of the number of biological parameters monitored, the analysis of the efficiency of a particular immunotherapy product in the stimulation of the immune response results in the necessity to analyze thousands of data (raw data).

Analysis of raw data can thus last for several months and be extremely tedious and expensive.

An aim of this invention is to provide a method for automating the treatment of raw results concerning measured expression levels and providing efficient reporting. Such a method increases the rapidity and reliability of the analyses, and reduces their overall cost. SUMMARY OF THE INVENTION

The present invention aims to solve problems related to the management of huge amounts of data, such as the raw data generated within the frame of preclinical or clinical studies, by providing a method for automatically processing multiple measurements of quantifiable biological parameters to obtain meaningful results, characterized in that it comprises the steps of:

(a) extracting a set of raw data including measurement values and annotations, said values comprising caliber-dependent values obtained under different caliber conditions and replicate values obtained under the same caliber condition ;

(b) from said raw data values and annotations and from related reliability range information, performing a correction process on said raw data including value correction and extended annotation generation reflecting abnormal values;

(c) performing on each group of caliber-dependent values of the same measurement after correction a best caliber value selection taking into account said extended annotations, thereby retaining sets of replicate values of the same measurements at best calibers;

(d) performing a mean value determination process on each set of replicate values, said determination process including an abnormal replicate value exclusion process; and (e) performing on said mean values a statistical and reporting process.

By "quantifiable biological parameter", it is meant any parameter that can be measured by use of a quantitative method such as relying on a calibration or standard curve and generating a numerical value (other than Boolean or logical) useful for monitoring a biological characteristic or activity, and preferably for monitoring an immune response. Such parameters notably include cytokines and chemokines expression levels, inflammatory or toxic markers, signalling molecules (such as phosphoproteins, kinases or apoptosis-related proteins), cytotoxic molecules (such as perforin, granzyme or FasL) expression levels, or specific antibodies concentrations.

The level and activity of such quantifiable biological parameters may be measured using various technologies available in the art including but not restricted to PCR, ELISA, quantitative flow cytometry measuring either optical density (OD), electroluminescence signals or fluorescence intensity, When the measurement is performed at the mRNA level (for instance using quantitative PCR), the results correspond to copy numbers of the tested mRNA. When the measurement is performed at the protein level, then intracellular or secreted concentrations of proteins are obtained.

The direct measurement of cytotoxicity may be performed using various technologies well known to those skilled in the art, and results in lysis percentage values. When the antibody response is monitored, titers of specific antibodies are obtained as raw data

In an advantageous embodiment, immune response parameters are selected from cytokines and chemokines expression levels (in particular cytokines or chemokines protein concentrations), inflammatory or toxic markers, signalling molecules (such as phosphoproteins, kinases or apoptosis-related proteins), cytotoxic molecules (such as perforin, granzyme or FasL) expression levels, lysis percentages corresponding to cytotoxic assays results, or specific antibodies concentrations. More advantageously, the immune parameters are cytokines and chemokines concentrations, preferably obtained using Luminex X-Map® technology.

The measurement of the biological parameters according to the method of the invention can be carried out on any type of biological material obtained from prokaryotic or eukaryotic organisms, such as microorganisms, plants, animals and humans. The method of the invention is preferably applied to body fluids of human organisms. Examples of such body fluids include without limitation blood, plasma, serum, spinal fluid, urine, faeces, lymph, saliva, sperm, and amniotic fluid. Moreover, the expression "biological material " also includes tissues from specific organs (such as brain tissue, muscle tissue, retinal tissue, kidney tissue, liver tissue, etc) taken from healthy organisms or organisms suffering from a disease state or disorder as well as cell culture, medium supernatants and cell lysates obtained from such cultures.

The present invention further provides a system for automatically processing multiple measurements of biological quantifiable parameters to obtain meaningful results, characterized in that it comprises:

^* an input for files of raw data including measurement values and annotations, said values comprising caliber-dependent values obtained under different caliber conditions and replicate values obtained under the same caliber condition ;

^* a processor connected to a data storage for:

- extracting said raw data from said files and for storing said raw data in said storage,

- performing on said raw data values and annotations, by using related reliability range information also contained in said storage, corrections on said raw data, said corrections including value correction and extended annotation generation reflecting abnormal values;

- performing on each group of caliber-dependent values of the same measurement after correction a best caliber value selection taking into account said extended annotations, thereby retaining sets of replicate values of the same measurements at best calibers;

- performing mean value determination on each set of replicate values, said determination including an abnormal replicate value exclusion; and

- generating from said mean values statistics and reporting; * an output for said statistics and reporting.

Preferred but non limiting features of the present invention are as follow:

- said step (a) comprises a step of inserting field delimiters into said raw data;

- said measurements values of quantifiable biological parameters are values derived from initial measurement values by using a transformation curve;

- said reliability range information is derived from a definition range of said transformation curve;

- said step (b) comprises an operation of testing measurement values against range limits, and replacing any value beyond a range limit by the corresponding limit value;

- extended annotation is generated by identifying the position of a value with respect to a set of range limits;

- said transformation curve is an interpolation curve whose equation is obtained by regression analysis;

- said initial measurements are fluorescence measurements;

- said fluorescence measurements are obtained by flow-cytometry;

- said fluorescence measurements are obtained by quantitative PCR or microarray analysis;

- said derived measurements values of quantifiable biological parameters are concentration values;

- said derived measurements values are cytokines and chemokines protein concentration values; - said caliber values are dilution factors;

- said abnormal replicate value exclusion process is performed on initial measurement values;

- abnormal replicate value exclusion process of step (d) comprises a Dixon's Test;

- said abnormal replicate value exclusion process of step (d) comprises calculating a Coefficient of Variation.

BRIEF DESCRIPTION OF THE DRAWINGS

The above and other objects, features and advantages of this invention will be apparent in the following detailed description of an illustrative embodiment thereof, with is to be read in connection with the accompanying drawings wherein:

- Figure 1 is a diagram representing the different steps of a method according to the present invention;

- Figure 2 is an example of an information file requested by the first step of a method according to the invention;

- Figure 3 is a example of an interpolation curve used to determine a range;

- Figure 4 is a diagram representing in detail the first part of the second step of a method according to the invention;

- Figure 5 is a diagram representing in detail the second part of the second step of a method according to the invention;

- Figure 6 is an example of a file obtained at the end of the second step of a method according to the invention;

- Figure 7 is a diagram representing in detail the third step of a method according to the invention;

- Figure 8a and 8b are diagrams representing two possible methods for realizing successive dilutions of a solution;

- Figure 9a-9c are three examples of replicate sets on which the method according to the invention is applied in accordance with their position on the interpolation curve; - Figure 10 is an example of a graph for a given patient and a given cytokine generated during the fourth step of a method according to the invention;

- Figure 1 1 is an example of a cluster for a given patient for a given stimulus generated during the fourth step of a method according to the invention.

DETAILED DESCRIPTION OF A PREFERRED EMBODIMENT

Referring to the drawings and initially to Figure 1 , a method according to the invention will now be described.

Input file 1 is in this example a file containing raw data obtained from acquisition software, named Bio-plex manager™, developed and commercially supplied by BioRad, Hercules, California, U.S.A. for the so- called X-Map® technology which is a multiparametric flow cytometry-based technology commercialized by Luminex Corporation, Austin, Texas, U.S.A. This system enables counting one by one particles, that are coupled to antibodies specific for various analytes (cytokines, chemokines, phosphoproteins, ....) and that can be identified by a specific color-code. Specific analytes are captured by the antibody-coupled beads and further revealed with a second analyte-specific antibody coupled with a fluorochrome. Beads are then counted and fluorescence intensity, related to the amount of analyte present in the sample, is measured on Luminex instrument. The BioRad data acquisition software generates a raw file containing both all the initial measurement data (fluorescence data in this case) obtained by such a system and the concentration data deduced from said initial measurement data.

However, various other methods for measuring concentrations and generating an input file 1 are known by the skilled person, and the invention is not limited to data processing of Luminex files. In particular, fluorescence raw data obtained after quantitative PCR or microarray analysis for instance, may also be automatically treated by the method according to the invention. In the following description, an illustrative embodiment is provided based on Luminex X-Map® data obtained for cytokines and chemokines measures. However, the same steps may be applied to other quantifiable biological parameters, in particular those listed above.

The method comprises four main successive parts, corresponding to the four major steps of the data processing which were previously manually done by lab workers. Preferably, this method is in the form of at least one script, which can be launched through a web page in XML format.

During a first step 100 of importing, tagging and compiling raw data, the script is fed with at least one file 1 , each file 1 corresponding to one 96- well plate of the Luminex lab instrument used. Indeed, samples in which expression measures are determined are generally organized by plates, each well of a plate being successively or simultaneously (depending on the technology used for measuring the expression level) analyzed by the Luminex instrument. The complete raw data obtained for the selected numbers of animals or patients, time points and parameters (cytokines and chemokines) generally correspond to such a high number of measures that it is generally obtained by using several plates.

All files 1 are converted into "tab delimited text" format (.txt) during the first operation 1 10. This format consists in a simple text format in which the content of each line of the original file is separated by a '\t' (tabulation) character. This conversion is useful because input files may come from different systems, and they have to be normalized before they can be inputted to the script.

Then the different files are compiled together in the second operation

120 of step 100. One plate can contain measures from different patients, different timepoints or different studies. Also, measures for one patient can be separated between two successive plates. The information file 2 contains all these information and the description of each plate. An example of this information file 2 can be seen in Figure 2. In this file, each line links the number of a plate with its contents. In the information file 2, the input files 1 are sorted in order to have the end of a plate corresponding with the beginning of the next plate. Then files are read line by line, indexed with related data from the information file 2, and stored in a global single file. One measurement is generally performed at least two or three times for increasing reliability of the test, and such multiple measures are named replicates. These replicates are counted when read.

At the end of the operation 120, a file 4 of raw results is obtained, preferably in the form of a spreadsheet such as an Excel® file. In this file 4, for each measurement there is a field "value" which contains the corresponding concentration value if existing. For this purpose, two search functions allow to determine for each cytokine or chemokine the range of concentrations and the lower and upper limit values for each plate. Once the data have been imported at step 100, data correction process of step b) is performed at step 200. In this regard, even if most of measurement values are reliable, they might contain artifacts (abnormal values). Moreover, in the case of Luminex flow-cytometry measures, as the concentrations are deduced from fluorescence initial measurements (MFI, Mean Fluorescence intensity), a concentration is obtained indirectly by interpolation from fluorescence according to a function 3 which typically is: ' ( 1 + exp{ 6(lo (a:- )— iogifi) } )) '

x is the concentration in pg/mL

f(x) is the fluorescence in MFI (Mean Fluorescence Intensity)

b, c, d, e and f are parameters.

In Figure 3, such function 3 is illustrated by the standard curve, with the estimated accuracy for each standard point (value in %). As coefficients b-f are calculated by regression for each plate, it is impossible to have a perfect match. In the case of the Luminex system, Bio-plex Manager performs the following calculation :

Calculated concentration / theoretical concentration x 100

In the case of a perfect match, this ratio is 100%. The user defines a range, generally 80%-120% or 70%-130%, named "recovery range". Values inside this range are estimated as close enough to theoretical values and considered as "in-range". Standard points at the extremities of the interpolation curve are often less accurate, so they can be excluded if they are outside the recovery range.

The software determines the set of points which are in the specified recovery range, ie the dynamic range of the curve. It corresponds to the area 31 . The excluded areas 32 and 33 of the figure 3 correspond to intervals in which values can still be calculated by function 3 but which are out of the specified limits. They are defined as "NIR" (Not In Range). However, not each software comprises statistical tools for calculating the dynamic range. In this case, this range is arbitrary defined as the definition range of the entire curve. There are no more areas NIR 32 and 33.

Out of the curve, there are two areas 34 and 35 in which values can still be obtained by extrapolation (i.e. calculated by continuation). Such values are appended with an asterisk in the field "value" in the raw data. Concentration values which can no longer be extrapolated (area 36 and 37) are marked as Out Of Range (OOR< or OOR> respectively, depending on whether they are under the lower standard point or above the upper standard point) in the field "value". A value can also be a blank or NA if data is missing or clearly aberrant for this measurement.

In the file 4 of raw results, each value which is not "in-range" is annotated " ^*** " in a field ConcR.

The process of step 200, as illustrated in detail in Figure 4 and Figure 5, aims to replace the current field value (which can be a number or a symbol) by two fields: one actual field "value", which can only contain a number, and a field "annotation" with contain information about said value. The corresponding data set is a Corrected Raw Results file 5. The various operations shown in the diagram of Figure 4 receive as input the raw results file 4 as well as lower and upper standard values of the dynamic range 31 defined from the curve, for a first processing 201 of every value.

These operations are as follows:

- if the value is normal (i.e. interpolated), this value is kept as it is and the annotation field is set as "OK";

- extrapolated values (i.e. those appended with an asterisk) are checked:

o if they are under lower limit, they are replaced by the lower limit value and annotated with "extraP_down"

o If they are above upper limit, they are replaced by the upper limit value and annotated with "extraP_up"

- "OOR<" and "OOR>" values are replaced respectively with the lower and upper limit values, and annotated respectively with "OOR<" and

"OOR>".

- If a value if missing, the value field is left blank and the annotation is set to "ND".

- If a value is aberrant, the value field is made blank and the annotation is set to Outlier".

A second data processing operation 202, represented in the figure 5 is then applied on concentration values taking into account both dilution and "concentration in range" (the field "ConcR") information. Thus, only values having an annotation " ^*** " are modified.

Lower and upper limit values of the range are multiplied by the dilution factor to obtain a shifted range. Each value which is not in range is replaced by the newly calculated lower or the upper limit of the range respectively if it is below or above the recovery range. Their annotations are replaced:

- Annotations of values which were "OK" are replaced by "NIR";

- Other annotations are left as such. An example of Corrected Raw Results file 5 can be observed in Figure 6. The range in this case for the cytokine TGF for example is 3.01 - 10353.99. The various annotations have been highlighted.

The Corrected Raw Results file 5 obtained is now ready to undergo a statistical analysis, which is the step 300 of the method according to the invention and which is shown in the diagram of Figure 7.

This step begins with an operation 310 which is the abnormal replicate value exclusion process of step d), ie detecting and discarding outlying replicates, which are replicates which are numerically too distant from the rest of the set.

For a question of statistical reliability, the values which are tested are the initial measured values (in our case MFI), and not converted concentration values, as the latter may have been corrected (replaced by lowest or highest limit of standard curve).

Practically, the vast majority of the sets of replicate do not contain an outlying value. Detection of outlying value is a classical problem in statistical processing, and numerous methods exist and are known by the skilled person. Preferably, a so-called Dixon's test, followed by a calculation of the Coefficient of Variation, will be used.

An outlying replicate is necessarily the minimum or the maximum of a set of replicates, and generally leads to a high value of the standard deviation. The basic principle of the Dixon's test is to sort the replicates, and calculate a Q threshold as follows: range

where gap is the absolute difference between the possible outlying replicate and the closest other replicate.

If Q is greater than a value Qtable (which is a constant depending of the number of replicates), the measurement can be considered as abnormal, and its value is replaced by NA (i.e. the measurement is removed). An additional filtering is advantageously applied to the Coefficient of Variation: the Coefficient of Variation value is first calculated by dividing the standard deviation of the replicates by their mean: c, =- °

μ

If the value C _v is greater than e.g. 50%, then the replicate is discarded as an outlying replicate, even if it was not significant after Dixon's test.

The next operation 320 (which can actually be performed simultaneously with the operation 310) aims to select the best caliber value for each set of replicates.

Indeed, it is well known that for measuring high concentrations, the solution can be diluted k times, and the obtained concentration multiplied by k. The caliber in the present case is the dilution factor: the fluorescence measurement system is accurate only in a limited range of intensity, and increasing the dilution factor allows to adapt this range. Generally, dilutions are realized "in cascade": in this example, samples are taken in each replicate of a set and separately diluted to obtain each replicate of the next dilution, and again (figure 8a). Such method required each replicate to be tested for finding the best dilution, due to a greater correlation between two dilutions of each replicate than between the three replicates at a given dilution.

However, if each set of replicates was coming from one unique primary solution independently diluted, it could be possible to select the best dilution for the set of replicates that contains the maximum available values, the mean being then calculated at this selected dilution. This is another method for realizing successive dilutions which is shown in the figure 8b. However, the first method is preferred because it could be easily automated.

In this example, an iterative process begins with the highest measurement values (obtained for the lowest dilution), and for each replicate, if extraP_up, OOR > or NIR appears, then the concentration value for that dilution is discarded and the next dilution is tested. The first concentration value which is not discarded is considered as the one obtained for the best dilution. If the concentration value is to be discarded for every dilution, then the sample has to be tested again at a higher dilution.

When the best dilution is found, a mean value (ref. 6) is calculated during the operation 330. This mean value is the mean of the accepted replicate values (outlying replicates having been removed) at the selected dilution.

The operation 340 is the final one of the statistical analysis in the present example. It aims to compare obtained values with reference values to measure the efficiency of stimulations.

These references values correspond to measurement values obtained by performing the same processing method for triplicates of samples which have not been stimulated. The obtained concentrations, referred to as medium values (ref. 7), of the medium for each cytokine or chemokine define control values.

Said medium values (ref. 7) are directly subtracted from the mean values (ref. 6) to obtain Delta values (variation caused by the stimulation).

The reliability of the obtained Delta Values can then be evaluated through the calculation of a Ratio and a Score.

A possible statistical estimator which produces good results for this application can be based on geometric mean value:

Δ = [mean] - [medium]

Ratio = [mean] /[medium]

In different embodiments, other known or specific estimators can be used.

Figures 9a-9c show three examples of replicates sets (represented on the interpolation curve 3) processed by step 300 of the method according to the invention. In the first example (9a), there is no outlying value, but the three replicates of the set are OOR>. It means that the dilution factor is too low, the next dilution has to be checked. The next dilution is a factor 5 dilution, and is the last dilution. The three replicates are ExtraP_up. So this is not the "best" dilution, but the "less bad" dilution. As there is no outlying value, the mean is calculated with the three values. However, each of their initial value was replaced by the upper limit because they are above dynamic range. The mean value is thus S5.

The second example (9b) is a normal case. Dixon's test and CV test are satisfying, and the first dilution factor appears to be the best (the three replicates are in dynamic range). Their mean is directly calculated.

The third example (9c) is a more complicate case. Here, replicates a(1 ) and b(1 ) are in-range. However c(1 ) is higher and ExtraP_up. It is discarded as an outlying value by the Dixon's test. The next dilution is tested for this out of range replicate. c(1/5) is now in the dynamic range, .the mean is calculated with a(1 ), b(1 ) and 5xc(1/5).

The last step 400 produces a report including graphic representation of results for a better understanding. Many types of diagrams are possible, but two in particular are preferred.

The cut-off values are the values obtained from reference samples analyses. Such values might be calculated and/or imported by the present invention. They are useful for biological interpretation.

The graph in the figure 10 aims to compare the efficiency of several stimulations for a given cytokine or chemokine and a given patient. There are boxes, one per stimulation. In each box is drawn a horizontal line which represents the cut-off threshold (ref. 8), and the ratio obtained for each replicate at each timepoint. Black dots represent individual values while vertical histogram bars (in green on the graph) represent mean values (ref. 7). Such diagram enables to easily compare stimulations through time.

The other preferred graph, which can be seen in figure 1 1 , is a cluster. For a given patient and a given stimulation, it aims to sort the pattern of soluble factors that behaved similarly at different time points (M-1 , D22, M6) for a given stimulation (NS3/1 ). The last column of this cluster represent the cut-off values, comparisons can be directly made.

It is to be understood that the invention is not limited to this particular embodiment, and that various changes and modifications may be brought by the one skilled in the art.

In particular, the embodiment of the present invention detailed below is based on Luminex X-Map® data corresponding to cytokines or chemokines protein concentrations.

However, as stated in the introduction, a quantifiable biological parameter may be monitored using alternative technologies. For instance, protein levels can be measured by classical ELISA (generating Optical Density values), by other multi-array technologies (such as MSD microarrays based on electroluminescence), or quantitative flow cytometry. Protein levels can sometimes be correlated with transcript (mRNA) levels that can be by quantitative PCR. However, when the expression level values of quantifiable biological parameter tested are obtained at the mRNA level instead of the protein level, the method according to the invention may be used in the same manner. In particular, the step of importing, tagging and compiling raw data is similar, there are also out of range values to predict and replace, and statistical analysis and reporting are similar.

Similarly, the same method might be applied to other quantifiable biological parameters such as inflammatory or toxic markers, signaling molecules, apoptosis-related proteins, cytotoxic molecules, or specific antibodies.