The invention relates to a method for in vitro differential diagnosis of lung adenocarcinoma (AC) versus squamous cell lung carcinoma (SCC), a method for in vitro differential diagnosis of squamous cell lung carcinoma (SCC) versus lung adenocarcinoma (AC), a panel of miRNA biomarkers, a panel of miRNA biomarkers for use for in vitro differential diagnosis of lung adenocarcinoma versus squamous cell lung carcinoma, a panel of miRNA biomarkers for use for in vitro differential diagnosis of squamous cell lung carcinoma versus lung adenocarcinoma, a use of the panel of miRNA biomarkers, and to a kit for in vitro differential diagnosis of lung adenocarcinoma versus squamous cell lung carcinoma and a kit for in vitro differential diagnosis of squamous cell lung carcinoma versus lung adenocarcinoma.

Field of the invention This invention generally relates to molecular clinical diagnostics, more specifically molecular differential diagnostics of histopathological subtypes of non-small cell lung cancer (NSCLC), in particular lung adenocarcinoma versus squamous cell carcinoma (SCC) by monitoring and analysing the expression of microRNAs (also referred to as miRNAs and mxRNome) in biological samples. Prior art

Lung cancer is the leading cause of death from malignant neoplasms worldwide. Nonsmall cell lung cancer is one of the most common types of lung cancer. Its two main histopathological subtypes are lung adenocarcinoma and squamous cell lung carcinoma. The incidence of adenocarcinoma and squamous cell carcinoma is as high as 85 to 90%. Proper diagnosis of the histopathological subtype of lung cancer allows for implementing an effective method of treatment and thus increases therapeutic benefits.

Diagnostic methods for stratification of subtypes of non-small cell lung cancer based on conventional histopathological tests are known. Often, however, results of such tests are equivocal or do not allow for a proper diagnosis of cancer histology. Now, a proper diagnosis is indeed crucial for the implementation of appropriate treatment and is a key element in qualifying patients for personalised therapies. Currently, if the diagnosis cannot be confirmed by the histopathological tests mentioned above, i.e. the type of tumour cannot be identified based on microscopic examination of the tissues collected, it is often necessary to perform surgery to obtain the definite confirmation of the cancer diagnosis and to plan appropriate treatment. Such surgeries are burdensome for the patient and often involve various complications.

Diagnostic assays for differential diagnosis of lung adenocarcinoma versus squamous cell lung carcinoma based on molecular methods are also known, i.e. analysis of expression of nucleic acid of one biomarker (Lebanon D, Benjamin H, Gilad S et al. Diagnostic assay based on hsa-miR-205 expression distinguishes squamous from nonsquamous non-small cell lung carcinoma. J Clin Oncol 2009; 27:2030-7, Bishop JA, Benjamin H, Cholakh H et al., Accurate classification of non-small cell lung carcinoma using a novel microRNA- based approach. Clin Cancer Res 2010; 16:610-19). It was found, however, that the results obtained with such assays based on the expression analysis of one biomarker are not sufficiently unequivocal and do not allow a definite distinction between adenocarcinoma and squamous cell lung carcinoma (Del Vescovo V, Cantaloni C, Cucino A, i wsp. miR-205 Expression levels in nonsmall cell lung cancer do not always distinguish adenocarcinomas from squamous cell carcinomas. Am J Surg Pathol 2011; 35:268-75).

Therefore, there are currently no diagnostic methods or means based on molecular diagnosing the lung cancer subtype that would allow overcoming the drawbacks known from the prior art. Developments in molecular diagnostics in this respect still leave room for improvement.

Brief description of the invention

The invention relates to a method for in vitro differential diagnosis of lung adenocarcinoma (AC) versus squamous cell lung carcinoma (SCC) in a subject, consisting in that:

(a) the expression level in a sample from the subject is determined quantitatively for the panel of miRNA biomarkers comprising: hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p, hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p, and

(b) the miRNA signature of the subject is identified by comparing quantitatively the expression level determined for the panel of biomarkers in the sample from the subject according to (a) with the expression of the panel of biomarkers in the reference sample being a prepared in a strictly controlled manner pool of solutions of RNAs isolated from AC and SCC tumours in equal amounts; wherein the expression level of biomarkers comprising hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p higher than in the said reference sample and the expression level of biomarkers comprising hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p lower than in the said reference sample indicate that the subject suffers from lung adenocarcinoma.

This invention further relates to a method for in vitro differential diagnosis of squamous cell lung carcinoma (SCC) versus lung adenocarcinoma (AC) in a subject, consisting in that:

(a) the expression level in a sample derived from the subject is determined quantitatively for the panel of miRNA biomarkers comprising the following miRNAs: hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p, hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p, and (b) the miRNA signature of the subject is identified by comparing quantitatively the expression level determined for the panel of biomarkers in the sample from the subject according to (a) with the expression of the panel of biomarkers in the reference sample being a mixture of solutions (pool) of RNAs isolated from AC and SCC tumours in equal amounts; wherein the expression level of biomarkers comprising hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p lower than in the said reference sample and the expression level of biomarkers comprising hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3pp higher than in the said reference sample indicate that the subject suffers from squamous cell lung carcinoma.

Preferably, in the methods for in vitro differential diagnosis according to the invention as a sample from a subject a sample selected from the group comprising fresh frozen tissue collected from the patient and formalin-fixed paraffin-embedded tissue is used. More preferably, the sample is a fresh frozen sample of lung tumour tissue. More preferably, the sample is a formalin-fixed paraffin-embedded sample of lung tumour tissue.

Preferably, in the methods according to the invention, the sample is derived from a subject being a human subject.

Preferably, in the methods for in vitro differential diagnosis according to the invention, the level of expression for the panel of miRNA biomarkers is determined using a method selected from next generation sequencing, hybridization, RT-PCR or microarray.

More preferably, in the methods for in vitro differential diagnosis according to the invention, the level of expression for the panel of miRNA biomarkers is determined using the next generation sequencing method.

Preferably, in the methods for in vitro differential diagnosis according to the invention the miRNA signature is identified using statistical or bioinformatic methods.

More preferably, in the methods according to the invention, the miRNA signature is identified using bioinformatic methods, in particular through data normalisation, transformation and distributional checks, differential abundance analysis, LASSO/elastic net regression, and correlation of abundances, even more preferably through normalisation of data obtained using the NGS method, transformation and distributional checks, differential abundance analysis and LASSO/elastic net regression, and correlation of abundances. The invention further relates to a panel of miRNA biomarkers comprising the following miRNAs: hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p, hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p.

Preferably, the panel of miRNA biomarkers according to the invention consists of the following miRNAs: hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p, hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p.

The invention also relates to a panel of miRNA biomarkers comprising: hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p, hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p for use in a method for in vitro differential diagnosis of lung adenocarcinoma versus squamous cell lung carcinoma in a subject, wherein the expression level of biomarkers comprising hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p higher than in the said reference sample and the expression level of biomarkers comprising hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p lower than in the said reference sample jointly indicate that the subject suffers from lung adenocarcinoma.

The invention also relates to a panel of miRNA biomarkers comprising the following miRNAs: hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p, hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p for use in a method for in vitro differential diagnosis of squamous cell lung carcinoma versus lung adenocarcinoma in a subject, wherein the expression level of biomarkers comprising hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p lower than in the said reference sample and the expression level of biomarkers comprising hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p higher than in the said reference sample jointly indicate that the subject suffers from squamous cell lung carcinoma.

The invention also relates to use of the panel of miRNA biomarkers as defined above for differential diagnosis of lung adenocarcinoma versus squamous cell lung carcinoma.

The invention also relates to use of the panel of miRNA biomarkers as defined above for differential diagnosis of squamous cell lung carcinoma versus lung adenocarcinoma. The invention also relates to a kit for differential diagnosis in vitro of lung adenocarcinoma versus squamous cell lung carcinoma, characterised in that it contains means to determine quantitatively the level of expression for the panel of miRNA biomarkers as defined above and an instruction to perform the differential diagnosis in vitro of lung adenocarcinoma as defined above. Preferably, the kit according to the invention comprises means for quantitatively determining the expression level of the panel of miRNA biomarkers using next generation sequencing method.

More preferably, the kit according to the invention comprises means for determining quantitatively the expression level of the panel of miRNA biomarkers consisting of hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p, hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p. The invention further relates to a kit for differential diagnosis in vitro of squamous cell lung carcinoma versus lung adenocarcinoma, characterised in that it comprises means for determining quantitatively the level of expression of the panel of miRNA biomarkers as defined above, and an instruction to perform a differential diagnosis in vitro of squamous cell lung carcinoma as defined above. Preferably, the kit for differential diagnosis in vitro of squamous cell lung carcinoma versus lung adenocarcinoma comprises means for determining quantitatively the level of expression of the panel of miRNA biomarkers using the next generation sequencing method.

More preferably, the kit for differential diagnosis in vitro of squamous cell lung carcinoma versus lung adenocarcinoma comprises means for determining quantitatively the level of expression of the panel of miRNA biomarkers consisting of the following: hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p, hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p.

Detailed description of the invention

The inventions according to the appended patent claims are useful for determination and differential diagnosis in vitro of the two main histopathological types of non-small cell lung cancer, namely non-squamous cell lung carcinoma, in particular adenocarcinoma, and squamous cell carcinoma. The methods according to the invention comprise isolating

RNA from a sample collected from a patient and quantitatively analysing the level of expression for the panel of selected miRNA biomarkers in the said sample. The increased expression of the biomarkers indicated above, and the reduced expression of other biomarkers indicated above compared to the reference sample indicated above, being a mixture of solutions of RNAs isolated from AC and SCC tumours in equal amounts, indicates a specific subtype of lung cancer. More precisely, the quantitative analysis of expression of said miRNA biomarkers allows for the determination of the tissue-specific expression profile of selected miRNA molecules, i.e. the so-called miRNA signature or differential miRNA expression signature, and for precise differentiation between the subtype of lung adenocarcinoma (AC) versus subtype of squamous cell lung carcinoma (SCC). The identification of the expression profile of said biomarkers, i.e. the miRNA signature, allows for obtaining a complete classification of lung cancer and for implementing appropriate treatment. In the methods according to the invention, as the sample from a patient a fresh frozen tissue or formalin-fixed paraffin-embedded tissue in the form of the so-called cytoblock may be used. Samples for use in the methods according to the invention may be derived from a material collected during a surgery or biopsy.

The methods according to the invention provide means for quantitatively determining and/or monitoring the expression of the panel of biomarkers at the nucleic acid level. Quantitative determination of the expression of the panel of miRNA biomarkers according to the invention may be performed by any method of quantitative determination of expression known in the art, preferably using the next generation sequencing (NGS). The NGS technology used for assessment of the miRNome profile, i.e. the level of tumour miRNA expression, enables precise assessment of the expression level of multiple miRNA molecules in multiple samples of biological material simultaneously. This accelerates analysis of multiple samples at the same time while lowering the cost of single analysis, which is crucial in clinical diagnosis. For the quantitative determination the expression level of miRNA the RT-PCR, hybridization or microarray methods and the like may also be used. Identification of the miRNA signature of a subject by way of quantitative comparison of the expression level determined for a panel of biomarkers in the sample from a subject with the expression of the panel of biomarkers in the said reference sample allows for obtaining a differential diagnosis of lung adenocarcinoma versus squamous cell lung carcinoma. miRNA signature may be identified by any method known in the art, such as statistical or bioinformatic methods, by way of the quantitative comparison of the expression level of a panel of biomarkers in a sample from a subject. According to the invention, the determination of the expression level of miRNA is particularly preferably performed using the NGS method, after which the miRNA signature is identified by bioinformatic methods as mentioned above. The panel of miRNA biomarkers according to the invention is a selected set of miRNAs. Such miRNAs from the panel of miRNA biomarkers according to the invention are selectively expressed on a higher or lower level in specific histopathological subtypes of lung cancer compared to the reference sample mentioned above, i.e. in non-squamous cell carcinomas, such as adenocarcinoma (AC), and in squamous cell carcinomas (SCCs). The panel of miRNA biomarkers according to the invention comprises the following miRNAs: hsa_miR_326, hsa_miR_450a_5p, hsa_miR__1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p, hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p.

For example, the use of the panel of miRNA biomarkers as described above allows for identifying the miRNA signature, which differentiates between lung adenocarcinomas versus squamous cell lung carcinomas. More precisely, in accordance with the invention a panel of miRNA biomarkers was identified, which biomarkers are specifically overexpressed in lung adenocarcinomas versus squamous cell lung carcinomas, i.e. whose expression level is higher than in the reference sample, as well as biomarkers, which are specifically underexpressed, i.e. whose expression level is lower than in the reference sample. This allows for the use of such panel in the differential diagnosis in vitro of lung adenocarcinoma versus squamous cell lung carcinoma, and in the differential diagnosis in vitro of squamous cell lung carcinoma versus lung adenocarcinoma.

Preferably, in accordance with the invention, the sample is a fresh frozen sample of lung tumour tissue or a formalin-fixed paraffin-embedded sample of lung tumour tissue. Preferably, such samples may be derived from a material collected during a surgery or biopsy.

In accordance with the invention the reference sample against which the expression of a panel of miRNA biomarkers is quantitatively compared is preferably a sample prepared in a strictly controlled manner consisting of a mixture of solutions of RNAs isolated from AC tumours and SCC tumours in equal amounts. According to the invention, the following miRNA biomarkers are specifically differentially overexpressed in lung adenocarcinoma versus squamous cell lung carcinoma: hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p.

According to the invention, the following miRNA biomarkers are specifically differentially underexpressed in lung adenocarcinoma versus squamous cell lung carcinoma: hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p.

In other words, this means that the following miRNAs are specifically differentially underexpressed in squamous cell lung carcinoma versus lung adenocarcinoma: hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p, and the following miRNAs are specifically differentially overexpressed in squamous cell lung carcinoma versus lung adenocarcinoma: hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p.

Accordingly, this allows for a differential diagnosis of squamous cell lung carcinoma (SCC) versus lung adenocarcinoma (AC), as the following miRNA biomarkers are specifically differentially underexpressed in squamous cell lung carcinoma versus lung adenocarcinoma: hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p, and the following miRNA biomarkers are specifically differentially overexpressed in squamous cell lung carcinoma versus lung adenocarcinoma: hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p.

Quantitative determination of the expression of the panel of biomarkers described above allows to identify the miRNA signature that reliably, clearly and reproducibly differentiates lung adenocarcinomas and squamous cell lung carcinomas. The present inventors conducted a clinical trial wherein were enrolled patients with diagnosed lung adenocarcinoma (AC) and squamous cell lung carcinoma (SCC). Following the quantitative determination of the panel of the selected miRNA biomarkers in each subject, they demonstrated that there was a statistically significant difference in the expression of the selected panel of miRNA biomarkers in AC subjects versus SCC subjects, based on the assumption that each analysed sample was compared against said reference sample. The reference sample, also referred to as the reference control, is used to normalise the expression levels of miRNA throughout the entire group of test samples, allowing for reliability of miRNA measurements in each sample regardless of tumour histology. This comparison system first involved a comparison of measurements of individual miRNAs for each sample against measurements of the same miRNAs in the reference sample. Then, the differences in expression between the individual samples and the reference sample were compared between groups in terms of AC versus SCC. The results showed that among the miRNAs identified, the expression level of twelve miRNAs was higher (upregulated) in the AC subjects versus the SCC subjects, while for five other miRNAs, the expression level was lower (downregulated) for AC versus SCC (see Table 4). Rather than the level of expression being higher or lower than for the reference sample, the crucial factor is the direction of change of the expression (upregulation or downregulation) in relation to the reference sample. Based on these results, it was surprisingly found that the differential analysis carried out on a panel of multiple miRNA biomarkers allows for obtaining a reliable miRNA signature allowing for reliable and unequivocal differential diagnosis in vitro of lung adenocarcinoma versus squamous cell lung carcinoma, as well as differential diagnosis in vitro of squamous cell carcinoma versus lung adenocarcinoma. In preferable embodiments according to the invention, the expression of the panel of miRNA biomarkers is determined quantitatively using the next generation sequencing. The following is a presentation of preferable methods of realisation of such sequencing. However, it must be recognised that the following preferable method of realisation is by no means intended to limit the scope of the invention as defined in the appended claims. I. Methodology of analytical measurements

1. Material dedicated to perform molecular analyses in accordance with the invention is either: • fresh tissue collected during surgery for NSCLC preserved in low temperature conditions (at least -80°C) (FF) or formalin-fixed and paraffin-embedded (paraffin block) (Formalin-Fixed Paraffin-Embedded, FFPE); or

• fresh tissue collected using biopsy techniques preserved in low temperature conditions (at least -80°C) or formalin-fixed paraffin-embedded (FFPE) in the form of the so-called cytoblock.

2. Histopathological verification

For histopathological verification, histopathological preparations are used to evaluate the number of cancer cells in the material prepared using conventional histopathological techniques, such as preparation of paraffin sections or frozen sections using cryostat, and H&E staining.

Moreover, an experienced pathologist performs a determination of the percentage of cancer cells in the preparation.

The minimum number of cancer cells required in the biological sample is 50%. 3. RNA extraction from tissue material

The total RNA from the sample from the test subject is then extracted using a commercial, fully optimised kit for isolation of total RNA with small RNA fraction dedicated for fresh tissue or FFPE-type material.

The efficiency of RNA extraction process is monitored using the so-called exogenous control being a mixture of synthetic 5'-phosphorylated miRNA molecules at various concentrations.

4. Qualitative and quantitative assessment of RNA solutions

The following qualitative and quantitative assessments are then carried out:

• assessment of solutions purity using the UV/VIS spectrophotometric method; · quantitative assessment of RNA using UV/VIS spectrophotometric method and fluorimetric technique;

• qualitative assessment of RNA in the context of the degree of integrity of RNA molecule by means of the microcapillary electrophoresis technique; wherein the minimum required RIN (RNA Integrity Number) value required for the RNA solution extracted from fresh material is 6.0, and the RIN value for RNA isolates deriving from FFPE-type material is not taken into account, as in this case the requirements involve determining the presence of small RNA fraction in the electrophorogram. 5. Preparation of cDNA libraries for next generation sequencing (NGS)

Next, cDNA libraries are prepared using a commercial fully optimized kit for preparing cDNA libraries for small RNA molecule template, based on amplification technique, compatible with Dlumina technology. 6. Qualitative assessment of cDNA libraries

The assessment of the structure and distribution of individual fractions of the cDNA library corresponding to individual fractions of molecules comprised within the small RNA fraction is performed using the microcapillary electrophoresis technique.

7. Library fractionation The size selection of cDNA products corresponding to the miRNA fraction is performed using the agarose gel electrophoresis technique.

8. Quantitative assessment of cDNA libraries after fractionation

After fractionation, an assessment of cDNA library concentration is performed using amplification technique. 9. Next generation sequencing (NGS)

Next generation sequencing is then performed on the HiSeq 4000 Illumina device.

The data obtained allow to identify the miRNA signature to differentiate lung adenocarcinoma versus squamous cell lung carcinoma.

According to the invention, such identification is preferably conducted using statistical methods, as known in the art, or bioinformatic methods. The following is a presentation of exemplary and preferable methods of realisation of bioinformatic methods. However, it must be recognised that the following preferable implementation method using bioinformatic methods is by no means intended to limit the scope of the invention as defined in the appended claims. II. Bioinformatic methodology

The analysis and identification of the differential miRNA expression signature comprises several steps. The first step involves aligning the filtered sequences reads to the reference genome using miRNA-specific parameters, and then counting the number of reads for each miRNA. The reference genome is a genome of a species to which a subject whose sample is tested belongs. Where said subject is a human being, the reference genome is the Homo sapiens genome. The quality of mapping and sample relations are studied using various methods and visualisation techniques. If samples of poor quality or data outliers are found, they should be excluded from further analysis at this step. The data are also normalised based on the library and miRNA lengths in order to reduce systematic noise caused by non-biological sources and to improve sample comparability. Fig. 14 shows preferable steps of the miRNA signature identification using bioinformatic methods according to this invention for the dataset obtained from quantitative determination of expression of the panel of miRNA biomarkers, preferably using miRNA sequencing. All analyses should be conducted using R language statistical calculation packages (e.g. R-related Bioconductor module).

1. Raw data

The reads should be obtained from the next generation sequencing device (e.g. HiSeq 4000 (Illumina)), e.g. based on the base counting method using manufacturer's software dedicated for this purpose.

2. Mapping

The reads should be aligned to the reference genome, preferably in the STAR 2.5.1b version, using the 2-pass alignment mode. The reference genome is a genome of a species from which a test sample of a subject is derived, as explained above. 3. Counting and normalisation

After alignment to the reference genome, the reads should be associated with the known miRNAs and the number of reads aligned within each miRNA should be counted. The data should then be normalised to remove variation between samples caused by non- biological reasons and to make the values comparable across the sample set. The reads should be normalised, e.g. using the TMM method. For statistical testing, the data are further subjected to log transformation, e.g. using the voom method.

4. Quality control

A quality control should be carried out to demonstrate the correlation between the replicates. It may be performed using the following methods: · visualisation of the miRNA expression distribution values across the sample set; • calculation of the minimum, median, mean values and maximum expression of miRNA of normalised samples;

• calculation of the correlation values between samples;

• hierarchical clustering the samples according to their general similarity;

• test sample relations using principal component analysis (PC A).

5. Statistical testing and data filtering

After all steps of preprocessing, statistical tests are carried out between the compared sample groups (AC versus (SCC) assuming that a reference sample is used which is a sample prepared in a strictly controlled manner consisting of a mixture of solutions of RNAs isolated from AC tumours and SCC tumours in equal amounts, to which all test samples are to be referred to. The results of statistical tests are then subjected to subsequent steps, of the so-called filtration of differentially expressed miRNA molecules. The filtering is based on statistical significance and the size of the difference in the mean expression levels between the sample groups. The fold changes of miRNA expression and the adjusted p-values, which are calculated using statistical testing, are used as filtering criteria. All miRNAs measurements are filtered in order to identify the molecules that demonstrate the strongest evidence for being differentially expressed between the compared groups.

The fold change (FC) describes the size of the difference in miRNA expression between the compared groups. The analysis is based on a linear modelling process carried out using the Limma package.

The p-value describes the reliability of the change in the miRNA expression values between the compared groups. In this analysis, the p-values used for filtering are the so- called modified t-test p-values or FDR (expected percentage of false discovery rates).

6. MiRNA signature identification

The process of identifying the expression signature based on miRNA sequencing data includes the following steps:

• Data Normalisation, Transformation and Distributional Checks;

• Differential Abundance Analysis and LASSO/Elastic-Net Regression; • Correlation of abundances.

7. Assessment of the diagnostic usefulness of the miRNA signature

Then, the assessment of diagnostic usefulness of the identified miRNA signatures is performed by generating a receiver operating characteristic curve and plotting true positive rates (TPRs) versus false positive rates (FPRs) at various threshold parameters. ROC curves may be generated using the MetaseqR Bioconductor package.

The miRNA signature of the subject obtained according to the invention for said panel of miRNA biomarkers according to the invention allows for a differential diagnosis of lung adenocarcinoma versus squamous cell lung carcinoma, when the determined expression level of the following miRNA biomarkers: hsa_miR_326, hsa_miR_450a_5p, hsa_miR_1287_5p, hsa_miR_556_5p, hsa_miR_542_3p, hsa_miR_30b_5p, hsa_miR_4728_3p, hsa_miR_450a_l_3p, hsa_miR_375, hsa_miR_147b, hsa_miR_7705, hsa_miR_653_3p in the sample from the test patient is higher than in the said reference sample, and the expression level of the following miRNA biomarkers: hsa_miR_944, hsa_miR_205_5p, hsa_miR_205_3p, hsa_miR_149_5p and hsa_miR_6510_3p in the sample from the test patient is lower than in the said reference sample.

This invention will now be illustrated in the following example and figures, which, however, are not intended to limit in any way the scope of the invention as defined in the claims.

Brief description of figures

Fig. 1 shows the sequence lengths distribution of the reads for the analysed samples.

Fig. 2 shows the curve describing the part of the miRNA expression values distribution for the analysed samples. Fig. 3 shows box plots for the miRNA expression values after normalisation.

Fig. 4 shows the Spearman’s correlation graph showing the level of correlation between the analysed samples.

Fig. 5 shows a dendrogram presenting the results of cluster analysis of all analysed samples.

Fig. 6 shows the results of the PC A analysis for the analysed samples. Fig. 7 shows the Volcano plot for comparison of AC and SCC.

Fig. 8 shows the MDA plot for comparison of AC and SCC.

Fig. 9 shows the PCA plot for comparison of AC and SCC.

Fig. 10 shows the results of LASSO regression analysis for the test data. Fig. 11 shows the results of LASSO regression analysis for cross-validation.

Fig. 12 shows the Heatmap presenting the miRNA expression signature for AC and SCC stratification.

Fig. 13 shows the analysis of the ROC curve for the identified miRNA signature differentiating lung adenocarcinomas and squamous cell lung carcinomas. Fig. 14 shows the analysis workflow for the expression level using bioinformatic methods for the dataset derived from miRNA sequencing.

Examples

All procedures and experimental analyses described below have been conducted using standard and commonly known methods used in the field to which this invention belongs, and commercially available test kits, reagents and equipment in accordance with the instructions of the manufacturers of the kits, reagents and equipment used, unless clearly indicated otherwise.

Example - Differential diagnosis of lung adenocarcinoma versus squamous cell lung carcinoma Materials and methods 1.1. Patients

The study was approved by the Ethics Committee of the Medical University in Bialystok. A declaration of informed consent for genetic testing has been obtained from each patient. The patients were recruited for the study in the frame of the MOB IT project. The study included 59 patients with diagnosed, surgical non-small cell lung cancer. The study inclusion criteria included the following: diagnosis of lung adenocarcinoma (AC) or squamous cell lung carcinoma (SCC) based on routine histopathological diagnosis; completely resected tumour (tumour-free resection margins); clinical stages I to IIIB; availability of representative fresh frozen tumour samples (biological material containing at least 50% tumour cells for RNA extraction) and no preoperative neoadjuvant chemotherapy. Detailed patient characteristics are presented in Table 1. SD, standard deviation

1.2. Histopathological diagnosis All lung tumour samples qualified for the study were histopathologically assessed using conventional histopathological techniques. The preparations were evaluated independently by two experienced pathologists. The histopathological diagnosis was made in line with the latest WHO lung cancer classification and IASLC/ATS/ERS International Multidisciplinary Classification of Lung Adenocarcinoma. In case of any doubt, the preparations were evaluated using immunohistochemical staining for the expression of thyroid transcription factor 1 (TTF-1) (immunohistochemical marker of adenocarcinoma) and for tumour protein p63 (p63) (squamous cell immunophenotype). Moreover, all tumour samples were assessed for the cancer cells percentage for RNA isolation. 2. RNA isolation and quality control

The total RNA along the small RNA fraction was extracted from fresh frozen tumour samples using a commercial RNA isolation kit - mirVana™ miRNA Isolation Kit (Ambion, USA) according to the manufacturer's protocol. The qualitative and quantitative assessment was performed using the spectrophotometric technique on the NanoDrop 2000c device (Thermo Scientific, USA). The concentration of RNAs solutions was also assessed using the fluorimetric technique on the Qubit device (Thermo Scientific, USA). The RNA integrity number (RIN) was determined using microcapillary electrophoresis technique on Bioanalyzer 2100 device (Agilent Technologies, USA). 3. Next generation sequencing (NGS) analysis

3.1. NGS libraries preparation

Preparation of cDNA libraries on the template of small RNA fraction molecules was performed using a commercial NEXTflex® Small RNA Sequencing Kit v3 device (gel- free & low input options) (BiooScientific, USA) compatible with Illumina technology according to the manufacturer's instructions.

3.2. Qualitative assessment of libraries

The assessment of the structure and distribution of individual library fractions, corresponding to individual fractions of molecules contained in the small RNA pool was carried out by means of the microcapillary electrophoresis technique using High Sensitivity DNA chips on Bioanalyzer 2100 Agilent device (Agilent Technologies, USA).

3.3. Fractionation of libraries

The selection of cDNA products of appropriate size (size selection) corresponding to the miRNA fraction was performed by means of the agarose gel electrophoresis technique using special gel cassettes on a Blue Pippin device (Sage Science, USA). 3.4. Quantitative assessment of cDNA libraries after fractionation

The concentration of cDNA libraries after fractionation was evaluated based on the amplification technique using a reagent kit, i.e. the KAPA Library Quantification Kit for Illumina Platforms (Roche, USA) according to the manufacturer’s protocol. 3.5. Next generation sequencing (NGS)

The sequencing of the prepared libraries was performed on the HiSeq 4000 (Illumina, USA) device in a standard manner according to the manufacturer’s protocol.

4. Bioinformatic analyses

All the analyses were performed using R language statistical calculation packages (R- related Bioconductor module).

4.1. Raw data

The reads were obtained from HiSeq 4000 (Illumina) device based on the base counting method using the manufacturer's software dedicated for this purpose.

4.2. Mapping

The reads were aligned to the Homo sapiens reference genome (Ensembl GROG 8 release) in STAR 2.5.1b version, using the 2-pass alignment mode according to the manufacturer’s protocol.

4.3. Counting and normalisation

After alignment to the reference genome, the reads were associated with the known miRNAs and the number of reads aligned within each miRNA was counted. The data were then normalised to remove variation between samples caused by non-biological reasons and to make the values comparable across the sample set. The reads were normalised using the TMM method. For statistical testing, the data were further subjected to log transformation using the voom method.

4.4. Quality control

Quality control was performed to show the correlation between replicates and to identify outliers. The following methods were used for quality control:

• visualisation of the miRNA expression values distribution across the sample set;

• calculation of the minimum, median, mean and maximum values of miRNA expression for normalised samples;

• calculation of the correlation of values between samples;

• hierarchical clustering the samples according to their general similarity; • test sample relations using principal component analysis (PCA).

4.5. Statistical testing and data filtering

After all data preprocessing steps described above, statistical tests were carried out between the compared sample groups (AC versus SCC) with the assumption of use of a reference sample which is a sample prepared in a strictly controlled manner consisting of a mixture of solutions of RNAs isolated from ACC tumours and SCC tumours in equal amounts, to which all test samples were referred to. The results from the statistical tests were then subjected to subsequent steps, of the so-called filtration of differentially expressed miRNA molecules. The filtering was based on the values of statistical significance and the size of difference in the mean expression levels between sample groups. The fold changes of miRNA expression and the adjusted p-values, which were calculated using statistical tests, were used as filtering criteria. All the measurements of miRNAs were filtered in order to identify the molecules that demonstrate the strongest evidence for being differentially expressed between the compared groups. Fold Change (FC) analysis was performed by linear modelling process using the Limma package. The p-values used for the filtering stage were the so-called modified t-test p-values or FDR (expected percentage of false discovery rates).

4.6. MiRNA signature identification

The process of identifying the signature based on miRNA sequencing data included the several steps as follows.

• Data Normalisation, Transformation and Distributional Checks;

• Differential Abundance Analysis and LASSO/Elastic-Net Regression.

4.7. Assessment of the diagnostic usefulness of the identified miRNA signature

The assessment of the diagnostic usefulness of the identified miRNA signature was assessed by generating a receiver operating characteristic (ROC) curve. The curve was plotted based on true positive rate (TPR) versus false positive value (FPR) at various threshold parameters. The MetaseqR Bioconductor package was used to generate the ROC curve. 5. Results

5.1. Raw data preprocessing

Raw sequence reads generated during the next generation sequencing on the HiSeq 4000 (Illumina) device have been counted by special base calling software coupled to the Illumina platform. The obtained reads were then subjected to an overall qualitative evaluation for each sample using the MultiQC v 1.7 modular tool, including i.a. the number of unique and duplicate reads, content of GC pairs, sequence lengths distribution, sequence duplication level, adapter content and sequence quality scores.

The reads were then trimmed in order to remove the adapter contamination and four random bases at both ends of the reads. The reads were evaluated again for quality using the MultiQC v 1.7 modular tool. Fig. 1 shows an exemplary qualitative parameter of the reads for the samples analysed, i.e. the sequence lengths distribution of the reads for the samples analysed.

The reads after preprocessing were aligned to the Homo sapiens reference genome (Ensembl GRCh38 release). A summary of alignment statistics of reads to the reference genome for some of the analysed samples is presented in Table 2.

Table 2. Characteristics of sequence reads after preliminary bioinformatic processing using the example of several biological samples.

The miRBase database annotations were used for mapping and counting the reads. The reads were linked to known miRNAs and the number of reads aligned for each miRNA was counted. The percentage of mapped reads differed between samples. The data were then normalised to remove variation between samples caused by nOn-biological reasons and to make the values comparable across the sample set. The reads were normalised using the TMM method. For statistical testing, the data were further subjected to log transformation using the voom method.

5.2. Quality control Quality control was performed to demonstrate the correlation between replicates and to identify outliers. The methods described below were used for quality control of the data analysed. Fig. 2 shows the expression values distribution across the sample set taking into account minimum, mean and maximum values and median expression of normalised samples as a curve describing a part of the distribution of miRNA expression values of the samples analysed. The expression values distributions across all the reads for all samples analysed are visualised by the box plot. Fig. 3 shows visualization of asinh- transformed expression values as a box plot for miRNA expression values after normalisation. Variation in the expression values distributions of the samples were compensated by means of normalisation, after which the miRNA expression values distributions of the samples were almost equal. The above procedures enabled a seamless comparative analysis of miRNA expression levels for all samples analysed. Correlation values between samples describe the similarity between samples on an overall level when all measurement characteristics of all samples are taken into account. In the next analysis the Spearman’s metrics were used, which metrics describe the similarity between the samples on a scale of 0-1. Value 0 denotes a complete lack of correlation between samples, while value 1 denotes an ideal correlation between samples. Fig. 4 shows the Spearman correlation graph for the samples showing the correlation level between the analysed samples.

Moreover, qualitative QC data were subjected to another analysis to further check the data structure. In the hierarchical cluster analysis, the samples were grouped according to their general similarity, taking into account all measurements of all samples. In said analysis, the samples were clustered using Euclidean metrics. The results of cluster analysis of all analysed samples were visualised as a dendrogram, which is branching out graph (Fig. 5). The dendrogram showed the most similar samples (in other words, the best correlating samples) found in the branches nearest to each other.

The sample relations have also been assessed using principal component analysis (PCA), which is an ordination technique complementary to clustering. This method allowed for grouping samples by their similarity. The results of the PCA analysis were visualized in three-dimensional space in Fig. 6.

5.3. Statistical testing and data filtering

After all steps of bioinformatic data preprocessing, statistical tests were carried out between the compared sample groups (AC versus SCC) with the assumption of use of a reference sample which is a sample prepared in a strictly controlled manner consisting of a mixture of solutions of RNAs isolated from ACC tumours and SCC tumours in equal amounts, to which all test samples were referred to. The results of statistical tests were then subjected to subsequent steps of the so-called filtration of differentially expressed miRNA molecules. The filtering was performed based on the values of statistical significance and the size of the difference in mean expression levels between sample groups. The fold changes of miRNA expression and the adjusted p-values, which were calculated using statistical tests, were used as filtering criteria. All measurements of miRNAs were filtered in order to identify the molecules that demonstrate the strongest evidence for being differentially expressed between the compared groups. Fold Change (FC) analysis was performed by linear modelling using the Limma package. The p-values used for the filtering step were the so-called modified t-test p-values or FDR (expected percentage of false discovery rates). The obtained results are shown in Table 3 and presented as diagrams in Figures 7, 8 and 9.

Table 3. A list of pre-identified miRNAs that show differentiated expression in AC and SCC groups.

5.4. miRNA signature identification

The process of miRNA signature identification based on miRNA sequencing data included the following steps.

In the first step, Differential Abundance Analysis and LASSO / Elastic-Net Regression were performed. The data were subjected to log2 transformation and a preliminary MDS plot was generated. Cut-off for the significant hit was FDR<= 0.05. Table 4 below lists the identified miRNAs showing differentiated expressions in the AC versus the SCC groups.

Table 4. A panel of identified miRNA biomarkers showing differentiated expression allowing for differential diagnosis of AC versus SCC.

LASSO regression results for the test data and cross-validation are presented in Figures 10 and 11.

In order to visualise the identified miRNA signature, differentiating between lung adenocarcinomas and squamous cell lung carcinomas, an appropriate heatmap showing the miRNA expression signature of AC and SCC stratification was plotted (Fig. 12).

5.5. Assessment of the diagnostic usefulness of the identified miRNA signature

The diagnostic usefulness of identified miRNA signature was assessed by generating a ROC curve. The curve was plotted based on true positive rate (TPR) against false positive value (FPR) at various threshold parameters. The MetaseqR Bioconductor package was used to generate the ROC curve. The tool allowed to generate the ROC curve using the matrix of p- values for the results of the differential miRNA expression analysis from the previous analysis given the vector based on direct observation for the differential expression and significance level. Fig. 13 shows the analysis of the ROC curve for the identified miRNA signature differentiating lung adenocarcinomas and squamous cell lung carcinomas.

6. Conclusions

The data obtained from the analyses above demonstrate that the identification of the miRNA expression signature is a precise tool to differentiate between the two main histopathological subtypes of NSCLC. The miRNomic signature is an excellent diagnostic model for the stratification of adenocarcinomas and squamous cell carcinomas in the context of the procedure of patient qualification for personalised therapies.