IN VITRO METHOD FOR THE DIAGNOSIS OF LUNG CANCER

Abstract

In vitro method for the diagnosis of lung cancer. The present invention is generally related to diagnostic assays. In particular, the present invention refers to the use of at least C4c fragment as diagnostic and prognostic lung cancer marker. C4c fragment can also be useful to estimate lung cancer risk and to decide whether a medical regimen has to be initiated and to determine whether the medical regimen initiated is efficient.

Claims

1. In vitro method for the diagnosis or screening of lung cancer in a subject which comprises: a. Determining the level of at least C4c fragment in a plasma sample isolated from the subject; and b. Comparing the C4c fragment level determined in step (a) with a reference control level of said C4c fragment, and c. Wherein if the C4c fragment level determined in step (a) is higher than the reference control level, it is indicative that the subject suffers from lung cancer.

2. In vitro method, according to claim 1, wherein indeterminate pulmonary nodules have been previously identified in the subject to be diagnosed and wherein if the C4c fragment level determined in step (a) is higher than the reference control level, it is indicative that the indeterminate pulmonary nodules identified in the subject are malignant.

3. In vitro method for deciding whether to administer a medical treatment to a subject suspected of suffering from lung cancer which comprises: a. Determining the level of at least C4c fragment in a plasma sample isolated from the subject; and b. Comparing the C4c fragment level determined in step (a) with a reference control level of said C4c fragment, and c. Wherein if the C4c fragment level determined in step (a) is higher than the reference control level, a medical treatment suitable for the treatment of lung cancer is selected to be administered to the patient.

4. In vitro method, according to claim 3, wherein indeterminate pulmonary nodules have been previously identified in the subject and wherein if the C4c fragment level determined in step (a) is higher than the reference control level, it is indicative that the pulmonary nodules identified in the subject are malignant and a medical treatment suitable for the treatment of lung cancer is selected to be administered to the patient.

5. In vitro method for determining the efficacy of a medical regimen in a patient already diagnosed with lung cancer which comprises: a. Measuring the level of at least C4c fragment in a plasma sample isolated from the patient prior to the administration of the medical regimen; and b. Measuring the level of said C4c fragment in a sample from the patient once started the administration of the medical regimen; and c. Comparing the C4c fragment levels measured in steps (a) and (b), in such a way that if the C4c fragment level measured in step (b) is lower than the C4c fragment level measured in step (a), it is indicative that the medical regimen is effective in the treatment of lung cancer.

6. In vitro method, according to claim 5, wherein malignant pulmonary nodules have been previously identified in the patient and wherein if the C4c fragment level measured in step (b) is lower than the C4c fragment level measured in step (a), it is indicative that the medical regimen is effective in the treatment of malignant pulmonary nodules.

7. In vitro method for determining the efficacy of a medical regimen in a patient already diagnosed with lung cancer comprising: a. Measuring the level of at least C4c fragment in a plasma from the patient once started the administration of the medical regimen; and b. Comparing the C4c fragment level measured in step (i) with a reference control level of the C4c fragment, and c. Wherein, if the C4c fragment level measured in step a) is not higher than the reference control level, it is indicative that the medical regimen is effective in the treatment of lung cancer.

8. In vitro method, according to claim 7, wherein malignant pulmonary nodules have been previously identified in the patient and wherein if the C4c fragment level measured in step a) is not higher than the reference control level, it is indicative that the medical regimen is effective in the treatment of malignant pulmonary nodules.

9. In vitro method, according to any of the previous claims, wherein the step a) comprises measuring the level of C4c fragment and prolactin, C4c fragment and CYFRA 21-1, C4c fragment and C-reactive protein (CRP), C4c fragment and prolactin and CYFRA 21-1, C4c fragment and prolactin and CRP, C4c fragment and CYFRA 21-1 and CRP, or C4c fragment and CYFRA 21-1 and CRP and prolactin.

10. In vitro method, according to any of the previous claims, wherein the lung cancer is selected from the group consisting of non-small cell lung cancer and small-cell lung carcinoma.

11. In vitro method, according to any of the previous claims, wherein the subject to be diagnosed or screened is an individual at high-risk for lung cancer.

12. In vitro method, according to any of the previous claims, characterized in that it is an immunoassay.

13. Use of at least C4c fragment in the in vitro diagnosis of lung cancer.

14. Use, according to claim 13, of at least C4c fragment for the in vitro diagnosis of lung cancer, wherein it is determined if previously identified indeterminate pulmonary nodules are malignant.

15. Use, according to any of the claim 13 or 14, wherein the following combinations of biomarkers are measured: C4c and prolactin, C4c and CYFRA 21-1, C4c and CRP, C4c and prolactin and CYFRA 21-1, C4c and prolactin and CRP, C4c and CYFRA 21-1 and CRP, or C4c and CYFRA 21-1 and CRP and prolactin.

16. Use, according to any of the claims 13 to 15, wherein the subject to be diagnosed or screened is an individual at high-risk for lung cancer.

Description

BRIEF DESCRIPTION OF THE FIGURES

[0028] FIG. 1. Structure of the complement component C4 and its proteolytic fragments.

[0029] FIG. 2. Quantification of C4d (A) and C4c (B) in plasma samples from early stage lung cancer patients and control individuals.

[0030] FIG. 3. Correlation between C4d and C4c plasma levels in both control individuals and early stage lung cancer patients (A), only control individuals (B) or only lung cancer patients (C).

[0031] FIG. 4. ROC curves obtained from the quantification of C4d (A) and C4c (B) in plasma samples from patients with early lung cancer and control individuals.

[0032] FIG. 5. Quantification of 24 cancer-related biomarkers in plasma samples from early stage lung cancer patients and control individuals measured by Luminex technology.

[0033] FIG. 6. Quantification of IL6 (A), prolactin (B), CYFRA 21-1 (C) and CRP in plasma samples from early stage lung cancer patients and control individuals using Cobas technology, and correlation of the levels of IL6 (A), prolactin (B) and CYFRA 21-1 (C) as measured using Luminex and Cobas technologies. Of note, no correlation is shown for CRP because this marker was not determined by Luminex technology.

[0034] FIG. 7. ROC curve obtained from the diagnostic model based on the quantification of C4c, prolactin, CYFRA 21-1 and CRP in plasma samples from patients with early lung cancer and control individuals.

[0035] FIG. 8. ROC curves from the quantification of C4c (A) and C4c/CYFRA 21-1/CRP (B) in plasma samples from patients with malignant and non-malignant pulmonary nodules.

[0036] FIG. 9. ROC curves for the discrimination between benign and malignant lung nodules obtained using a diagnostic model based on C4c quantification in plasma samples and the clinical features age, smoking status, smoking history and nodule size (A), or based on the quantification of C4c, CYFRA 21-1 and CRP in plasma samples and the same clinical features (B).

[0037] FIG. 10. ROC curves for the discrimination between benign and malignant lung nodules obtained using the Gould's diagnostic model (A), a model based on the Gould's model and C4c quantification in plasma (B), or a model based on the Gould's model and the quantification of C4c, CYFRA21-1 and CRP in plasma (C).

[0038] FIG. 11. ROC curves obtained from the diagnostic model based on the quantification of C4c (A), C4c/CYFRA (B) and C4c/CYFRA/CRP (C) in plasma samples from the screening of individuals at high-risk for lung cancer.

DETAILED DESCRIPTION OF THE INVENTION

Example 1. C4c has a Better Performance than C4d in the Diagnosis of Lung Cancer.

[0039] Description of the Experiment

[0040] C4c and C4d-containing fragments (from now on referred as C4d) were determined in plasma samples from early stage lung cancer patients and control individuals.

[0041] Material and Methods

[0042] Plasma samples from 39 patients with surgically resectable lung cancer at early stages (I and II) and from 39 healthy people (matched by sex, age and smoking history) were obtained at the Clinica Universidad de Navarra. A summary of the characteristics of these patients and controls is shown in Table 1. Lung tumors were classified according to the World Health Organization 2004 classification.

TABLE-US-00001 TABLE 1 Epidemiological and clinical characteristics of cases and controls. Control Lung cancer Characteristics individuals patients Sex Male 34 34 Female 5 5 Age 70 years 30 31 >70 years 9 8 Smoking status Ex-smoker* 27 26 Current smoker 12 13 Histology Adenocarcinoma 19 Squamous cell carcinoma 20 Nodule size 3 cm 22 >3 cm 17 Stage I 31 II 8 *Also including one never smoker

[0043] Plasma samples were spun down at 300 g for 10 min, and the supernatants were collected. Samples were stored at 80 C. until analysis. C4d was determined using a commercially-available enzyme-linked immunosorbent assay (A008, Quidel). The assay recognizes all C4d-containing fragments of activated C4 (including C4b, iC4b and/or C4d). C4d-containing fragments of activated C4 are referred in this document as C4d. C4c levels were determined as previously described and expressed as arbitrary units [Pilely K, Skjoedt M O, Nielsen C, Andersen T E, Louise Aabom A, Vitved L, Koch C, Skjdt K, Palarasah Y. A specific assay for quantification of human C4c by use of an anti-C4c monoclonal antibody. J Immunol Methods 2014; 405: 87-96].

[0044] Receiver operating characteristic (ROC) curves were generated in order to evaluate the diagnostic performance of the biomarkers. Other parameters such as sensitivity, specificity, positive predictive value, negative predictive value, likelihood positive ratio and likelihood negative ratio were also assessed. Statistical analyses were carried out with STATA/IC 12.1.

[0045] Results

[0046] Levels of C4d in plasma samples from control individuals and lung cancer patients were 0.790.28 g/ml vs. 1.000.46 g/ml, respectively (FIG. 2A). In the case of C4c, plasma levels were 8830 AU in control individuals, and 16460 AU in lung cancer patients (FIG. 2B). Associations between the levels of these two molecular markers and sex, age, smoking status, histology, nodule size and stage are shown in Table 2 and Table 3.

TABLE-US-00002 TABLE 2 Association between C4d plasma levels and clinicopathological features of lung cancer cases and controls. Controls Cases Characteristics C4d (g/ml) p value.sup.1 C4d (g/ml) p value.sup.1 Sex Male 0.78 0.30 0.737 1.03 0.47 0.172 Female 0.77 0.09 0.79 0.11 Age (years) 70 0.75 0.28 0.257 1.02 0.48 0.767 >70 0.89 0.30 0.91 0.24 Smoking status Ex-smoker 0.73 0.24 0.361 1.04 0.50 0.290 Current smoker 0.88 0.35 0.91 0.32 Histology Adenocarcinoma 0.88 0.20 0.221 Squamous cell carcinoma 1.11 0.57 Nodule size 3 cm 0.90 0.29 0.130 >3 cm 1.13 0.58 Stage I 0.93 0.34 0.044 II 1.28 0.69 .sup.1Mann-Whitney U test

TABLE-US-00003 TABLE 3 Association between C4c plasma levels and clinicopathological features of lung cancer cases and controls. Controls Cases Characteristics C4c (AU).sup.1 p value.sup.2 C4c (AU).sup.1 p value.sup.2 Sex Male 89 31 0.801 166 63 0.614 Female 83 21 154 34 Age (years) 70 84 28 0.243 162 61 0.835 >70 100 32 170 58 Smoking status Ex-smoker 87 30 0.831 162 65 0.882 Current smoker 91 30 167 52 Histology Adenocarcinoma 157 54 0.431 Squamous cell carcinoma 171 66 Nodule size 3 cm 155 65 0.223 >3 cm 176 53 Stage I 159 58 0.251 II 184 66 .sup.1AU: Arbitrary units. .sup.2Mann-Whitney U test

[0047] Correlation studies showed that the quantification of C4d and the quantification of C4c were not equivalent. Thus, only a weak association was observed between both markers when all samples were analyzed together (FIG. 3A). Moreover, this association disappeared when controls and cases were analyzed separately (FIG. 3B-C).

[0048] To evaluate the capacity of the markers to discriminate between cases and controls, ROC curves were generated (FIG. 4). Areas under the ROC curve were 0.69 (95% CI=0.57-0.81) for C4d and 0.86 (95% CI=0.77-0.95) for C4c. The area under the curve for C4c was significantly better than that for C4d (p=0.001). Likewise, the performance of C4c was superior to that of C4d in all other diagnostic characteristics analyzed (Table 4).

TABLE-US-00004 TABLE 4 Diagnostic performance of the determination of C4d and C4c levels in plasma samples from early stage lung cancer patients. C4d C4c Sensitivity 51% 78% Specificity 67% 95% Positive predictive value 61% 93% Negative predictive value 58% 77% Likelihood positive ratio 1.54 14.00 Likelihood negative ratio 0.73 0.30 Correctly classified 59% 83%

[0049] Conclusion

[0050] This analysis evidences that plasma samples taken from patients with lung cancer at early stages (I and II) contain higher levels of C4c than plasma samples taken from control individuals. Moreover, the determination of this biomarker has a significantly superior diagnostic performance than the determination of C4d, another proteolytic fragment derived from complement activation previously proposed as a diagnostic marker for lung cancer.

Example 2. Combination of C4c with Other Cancer Markers Increases its Diagnostic Potential.

[0051] Description of the Experiment

[0052] The diagnostic performance of several cancer markers was evaluated in the set of patients described in Example 1. The diagnostic information provided by the quantification of some of these markers was used to improve the diagnostic accuracy of C4c.

[0053] Material and Methods

[0054] Epidemiological and clinical characteristics of the early stage lung cancer patients are described in Example 1. The analytical evaluation of the cancer markers was performed in plasma samples using the Human Circulating Cancer Biomarkers Magnetic Bead Panel 1 (HCCBP1MAG-58K, Millipore) with Luminex technology. The markers analyzed were: AFP, total PSA, CA15-3, CA19-9, MIF, TRAIL, leptin, IL-6, sFasL, CEA, CA125, IL-8, HGF, sFas, TNF, prolactin, SCF, CYFRA 21-1, OPN, FGF2, -HCG, HE4, TGF, VEGF. A Cobas analyzer (Roche) was also used for the determination of IL6, prolactin, CYFRA 21-1 and CRP. Logistic regression was used to generate the diagnostic models.

[0055] Results

[0056] Results from the evaluation of the 24 potential diagnostic markers with Luminex technology are shown in Table 5 and FIG. 5. Significant differences between controls and cases were found in the plasma levels of IL6, prolactin and CYFRA 21-1.

TABLE-US-00005 TABLE 5 Marker levels (mean SD) in plasma samples from control individuals and lung cancer patients at early stages, as determined by Luminex technology. Marker Controls Cases p value.sup.1 AFP (pg/ml) 5509 23409 1696 951 0.893 Total PSA (pg/ml) 956 1174 714 888 0.475 CA15-3 (U/ml) 13.3 11.7 11.2 6.5 0.549 CA19-9 (U/ml) 21.0 9.8 20.5 9.8 0.877 MIF (pg/ml) 1526 1002 1550 2314 0.124 TRAIL (pg/ml) 183 102 168 63 0.916 Leptin (pg/ml) 25304 22950 24553 16283 0.586 IL6 (pg/ml) 11.8 27.4 6.9 4.1 0.022 sFASL (pg/ml) 56.3 146 21.3 14.5 0.455 CEA (pg/ml) 1326 986 3382 9251 0.738 CA125 (U/ml) 11.1 25.0 6.2 3.5 0.319 IL8 (pg/ml) 5.8 5.6 8.4 11.3 0.254 HGF (pg/ml) 300 365 228 53 0.340 sFAS (pg/ml) 2273 1018 2054 855 0.463 TNF (pg/ml) 10.4 9.1 9.1 2.4 0.996 Prolactin (pg/ml) 8759 5885 16442 18237 0.032 SCF (pg/ml) 59.8 52.4 54.0 19.1 0.569 CYFRA 21-1 (pg/ml) 1294 537 3797 5529 <0.001 OPN (pg/ml) 50181 25313 59616 31881 0.310 FGF2 (pg/ml) 162 140 138 48.9 0.798 bHCG (mU/ml) 0.32 1.2 0.09 0.26 0.864 HE4 (pg/ml) 7148 11785 5036 2303 0.860 TGF (pg/ml) 106 597 1.5 4.2 0.890 VEGF (pg/ml) 74.9 231 40 106 0.847 .sup.1Mann-Whitney U test

[0057] We next validated the results obtained for IL6, prolactin and CYFRA 21-1 using Cobas technology. Of note, CYFRA 21-1 could not be determined in one control sample. Correlations between Luminex and Cobas technologies were also evaluated for each of these markers. Additionally, we analyzed the levels of CRP, a biomarker that was not present in the Millipore panel. Results of these analyses are showed in Table 6 and FIG. 6. All markers showed significant differences between cases and controls.

TABLE-US-00006 TABLE 6 Marker levels (mean SD) in plasma samples from control individuals and lung cancer patients at early stages, as determined by Cobas technology. Marker Controls Cases p value.sup.1 IL6 (pg/ml) 6.7 10.4 9.5 8.3 0.015 Prolactin (pg/ml) 8330 5970 15389 18385 0.003 CYFRA 21-1 (pg/ml) 1294 537 3797 5529 <0.001 CRP (pg/ml) 2803 3844 14692 24412 0.004 .sup.1Mann-Whitney U test

[0058] The next objective was to develop diagnostic models using combinations of the levels of C4c and these four markers (IL6, prolactin, CYFRA 21-1 and CRP). For that, in first place, a univariate analysis was performed for each of the markers to determine their individual predictive capacity. As shown in Table 7, IL6 did not predicted malignancy and, therefore, was not included in the multivariate analysis.

TABLE-US-00007 TABLE 7 Simple logistic regression for the evaluation of C4c, prolactin, CYFRA 21-1 and CRP as potential diagnostic markers in lung cancer. Marker LR chi.sup.2 p value C4c 39.18 <0.001 IL6 1.80 0.180 Prolactin 7.40 0.006 CYFRA 21-1 24.38 <0.001 CRP 14.52 <0.001 LR chi.sup.2: likelihood ratio chi-square test

[0059] A multivariate model was developed using the plasma levels of C4c, prolactin, CYFRA 21-1 and CRP. Of note, the individual in whom CYFRA 21-1 could not be determined was removed from the multivariate analysis. Multivariate logistic regression analysis generated a model with a value of 52.09 for the likelihood ratio chi-square test (p<0.001). The predicted probabilities of the model were compared with the final diagnoses, and a ROC curve was constructed. The area under the ROC curve was 0.91 (95% CI=0.83-0.98) (FIG. 7). Other diagnostic characteristics of the model are shown in Table 7. This multivariate diagnostic model including the four markers was able to correctly classify a higher percentage of subjects than C4c alone (87% vs 83%).

TABLE-US-00008 TABLE 7 Performance of the lung cancer diagnostic model based on the plasma levels of C4c, prolactin, CYFRA 21-1 and CRP. C4c/Prolactin/CYFRA 21-1/CRP Sensitivity 82% Specificity 92% Positive predictive value 91% Negative predictive value 83% Likelihood positive ratio 10.39 Likelihood negative ratio 0.19 Correctly classified 87%

[0060] Associations between the predictive probabilities of malignancy derived from the model and clinicopathological characteristics of the patients and controls are shown in Table 8.

TABLE-US-00009 TABLE 8 Association between the probabilities obtained from the multivariate model, based on C4c/Prolactin/CYFRA 21-1/CRP plasma levels, and the characteristics of lung cancer patients and controls. Controls Cases Characteristics Score p value.sup.1 Score p value.sup.1 Sex Male 0.23 0.19 0.915 0.78 0.32 0.166 Female 0.22 0.16 0.76 0.17 Age (years) 65 0.18 0.19 0.015 0.74 0.32 0.237 >65 0.28 0.17 0.82 0.28 Smoking status Ex-smoker 0.23 0.19 0.730 0.79 0.30 0.766 Current smoker 0.22 0.19 0.77 0.31 Histology Adenocarcinoma 0.69 0.31 0.012 Squamous cell carcinoma 0.86 0.26 Nodule size 3 cm 0.70 0.33 0.019 >3 cm 0.89 0.22 Stage I 0.76 0.31 0.186 II 0.86 0.25 .sup.1Mann-Whitney U test

[0061] Multivariate models based on C4c and different combinations of the other three markers were also generated. Table 9 shows areas under the ROC curves and percentages of correctly classified events from these combinations. In most cases, the combined models improved the diagnostic characteristics of C4c alone.

TABLE-US-00010 TABLE 9 C4c-based diagnostic models generated by logistic regression using different combinations of C4c and prolactin, CYFRA 21-1 and/or CRP. Correctly Markers in the model Area under the curve classified events C4c 0.87 (95% CI = 0.77-0.95) 83% C4c/Prolactin 0.87 (95% CI = 0.79-0.95) 82% C4c/CYFRA 21-1 0.91 (95% CI = 0.84-0.98) 87% C4c/CRP 0.87 (95% CI = 0.78-0.95) 85% C4c/Prolactin/ 0.90 (95% CI = 0.83-0.98) 87% CYFRA 21-1 C4c/Prolactin/CRP 0.87 (95% CI = 0.79-0.95) 85% C4c/CYFRA 21-1/CRP 0.91 (95% CI = 0.83-0.98) 87%

[0062] Conclusion

[0063] These analyses evidence the capacity to diagnose lung cancer of models based on the combination of plasma levels of C4c with prolactin and/or CYFRA 21-1 and/or CRP.

Example 3. Determination of C4c in Combination with Other Protein Markers can be Used to Discriminate Benign from Malignant Indeterminate Pulmonary Nodules.

[0064] Description of the Experiment

[0065] The capacity of the C4c-based models to discriminate between patients with and without lung cancer was evaluated in a set of plasma samples from patients presenting benign or malignant lung nodules discovered by chest CT.

[0066] Material and Methods

[0067] A set of plasma samples from Vanderbilt University Medical Center was used in the study. This cohort included plasma samples from 138 patients presenting indeterminate lung nodules discovered by chest CT. Lung nodules were defined as rounded opacities completely surrounded by lung parenchyma. Seventy six indeterminate lung nodules were diagnosed as lung cancers by pathological examination, whereas the remaining 62 were diagnosed as non-malignant. Clinical features of malignant and non-malignant nodules are shown in Table 10. Diagnosis of non-malignant nodules included lung lesions such as chronic obstructive pulmonary disease, emphysema, inflammatory disease, granulomatous lesions, and hamartomas.

TABLE-US-00011 TABLE 10 Clinical and epidemiological features in the set of patients with indeterminate pulmonary nodules. Non-malignant Malignant- Characteristics lung nodules lung nodules Sex Male 36 50 Female 26 26 Age 65 46 33 >65 16 43 Smoking status Never 16 2 Former 24 42 Current 22 32 Pack-years 50 45 35 >50 17 41 Nodule size 3 cm 46 30 >3 cm 12 46 Not available 4 FEV1% predicted 80 29 50 >80 17 12 Not available 16 Histology.sup.1 ADC 26 SCC 16 LCC 6 SCLC 15 NSCLC NOS 13 Stage I-II 17 III-IV 39 Not available 20 Status Alive 23 Death 53 .sup.1ADC: Adenocarcinoma; SCC: Squamous cell carcinoma; LCC: Large cell carcinoma; SCLC: Small cell carcinoma; NSCLC NOS: Non-small cell lung cancer not otherwise specified.

[0068] Prolactin, CYFRA 21-1 and CRP plasma levels were analyzed using Cobas technology (Roche). C4d was evaluated as indicated in Example 1. C4c was evaluated using an enzyme-linked immunosorbent assay. Briefly, 96 well plates were coated with 200 ng of capture antibody (anti-C4c antibody; ref. CAM072-18, BioPorto Diagnostics). Plates were washed with wash buffer (PBS, 0.05% Tween-20, pH=7.4), and blocked with the same buffer. Samples were diluted 1:8 in assay buffer (wash buffer containing 20 mM EDTA). A plasma sample from a healthy individual, diluted 1:2, 1:4, 1:8 and 1:16, was used as a reference for quantification. After 1 hour at room temperature, plates were washed and a detection antibody was added (1:50,000; biotinylated antiC4b antibody; ref. HYB162-02B, BioPorto Diagnostics). Plates were washed and developed with streptavidin (1:200) and Substrate Reagent Pack (ref. DY999; R&D Systems). Results were calculated as relative values to those found in the reference sample and expressed as arbitrary units (AU). Logistic regression was used to generate the diagnostic models.

[0069] Results

[0070] Plasma C4c levels were significantly higher in lung cancer patients than in their non-lung cancer counterparts (9.893.95 vs 7.175.16 AU; p=0.002). The area under the ROC curve was 0.66 (95% CI=0.56 to 0.75) (FIG. 8A). In the case of C4d, there were no statistical differences in the plasma levels of this marker between individuals with benign and malignant nodules (4.602.06 g/ml vs. 5.092.10 g/ml; p=0.212), suggesting that the determination of C4d-containing fragments of activated C4 is ineffective in this clinical context. The diagnostic performance of C4c is summarized in Table 11.

TABLE-US-00012 TABLE 11 Performance of the determination of C4c levels in plasma samples as a potential diagnostic marker for the discrimination between benign and malignant pulmonary nodules. C4c Sensitivity 88% Specificity 44% Positive predictive value 6% Negative predictive value 98.9% Likelihood positive ratio 1.56 Likelihood negative ratio 0.27 Correctly classified 68% .sup.1Positive and negative predictive values were calculated with an estimated prevalence of malignant nodules in this clinical setting of 4%.

[0071] Associations between C4c levels and epidemiological and clinical characteristics of patients are shown in Table 12.

TABLE-US-00013 TABLE 12 Association between C4c plasma levels and clinical features in the set of patients with indeterminate pulmonary nodules. Benign Malignant Characteristics C4c (AU).sup.1 p value.sup.2 C4c (AU).sup.1 p value.sup.2 Sex Male 6.72 4.12 0.663 9.50 3.17 0.206 Female 7.78 6.36 10.68 5.11 Age (years) 65 8.11 5.46 0.398 10.70 4.79 0.753 >65 6.76 4.02 9.45 3.37 Smoking status Never smoker 9.13 6.80 0.212 11.57 1.04 0.364 Former/current smoker 6.48 4.34 9.85 3.99 Pack-years 50 7.08 5.59 0.795 9.38 3.40 0.823 >50 7.39 3.64 10.34 4.36 Nodule size 3 cm 7.76 5.21 0.219 9.71 3.37 0.807 >3 cm 6.02 4.93 10.01 4.32 FEV1% predicted 80 5.96 4.13 0.700 10.08 4.28 0.430 >80 6.44 7.29 10.43 3.96 Histology.sup.3 ADC 11.55 4.25 <0.001 SCC 7.51 2.87 LCC 12.52 4.74 SCLC 8.04 2.70 NSCLC NOS 10.45 3.18 Stage I-II 11.32 4.88 0.964 III-IV 10.25 3.84 Status Alive 10.96 4.19 0.442 Death 9.43 3.79 .sup.1Arbitrary units. .sup.2Mann-Whitney U test or Kruskal Wallis test. .sup.3ADC: Adenocarcinoma; SCC: Squamous cell carcinoma; LCC: Large cell carcinoma; SCLC: Small cell carcinoma; NSCLC NOS: Non-small cell lung cancer not otherwise specified.

[0072] The next objective was to develop diagnostic models using the combined information provided by C4c and the three protein markers found differentially expressed between cases and controls in Example 2 (prolactin, CYFRA 21-1 and CRP). A logistic regression analysis performed for each of the markers individually is shown in Table 13.

TABLE-US-00014 TABLE 13 Univariate logistic regression for the evaluation of C4c, prolactin, CYFRA 21-1 and CRP as potential diagnostic markers for the discrimination between benign and malignant pulmonary nodules. Marker LR chi.sup.2 p value C4c 12.11 <0.001 Prolactin 2.54 0.111 CYFRA 21-1 45.12 <0.001 CRP 19.87 <0.001

[0073] Based on the univariate analyses, C4c, CYFRA 21-1 and CRP were predictors of malignancy and were included in the multivariate logistic regression analysis. This study generated a model with a value of 64.04 for the likelihood ratio chi-square test (p<0.001). The predicted probabilities of the model were compared with the final diagnoses, and a ROC curve was constructed. The area under the curve was 0.86 (95% CI=0.80-0.92) (FIG. 8B). Other diagnostic characteristics of the model are shown in Table 14.

TABLE-US-00015 TABLE 14 Performance of the diagnostic model for the discrimination between benign and malignant pulmonary nodules based on plasma levels of C4c, CYFRA 21-1 and CRP. C4c/CYFRA 21-1/CRP Sensitivity 75% Specificity 85% Positive predictive value.sup.1 18% Negative predictive value.sup.1 98.8% Likelihood positive ratio 5.17 Likelihood negative ratio 0.29 Correctly classified 80% .sup.1Positive and negative predictive values were calculated with an estimated prevalence of malignant nodules in this clinical setting of 4%.

[0074] Associations between the predictive capacity of the model and characteristics of the patients are shown in Table 15. Interestingly, the predictive probabilities of malignancy were significantly associated with nodule size and vital status in the malignant group.

TABLE-US-00016 TABLE 15 Association between the scores obtained from the multivariate model, based on C4c/CYFRA 21-1/CRP plasma levels, and the characteristics of patients with malignant lung nodules and controls with benign nodules. Benign Malignant Diagnostic Diagnostic Characteristics score (AU).sup.1 p value.sup.2 score (AU).sup.1 p value.sup.2 Sex Male 0.33 0.19 0.887 0.74 0.27 0.466 Female 0.34 0.21 0.70 0.28 Age (years) 65 0.33 0.20 0.987 0.74 0.28 0.509 >65 0.34 0.21 0.71 0.26 Smoking status Never smoker 0.40 0.21 0.111 0.65 0.04 0.436 Former/current smoker 0.31 0.19 0.73 0.27 Pack-years 50 0.33 0.19 0.994 0.70 0.25 0.265 >50 0.34 0.22 0.75 0.28 FEV1% predicted 80 0.32 0.18 0.309 0.72 0.28 0.178 >80 0.27 0.20 0.64 0.20 Nodule size 3 cm 0.34 0.20 0.863 0.58 0.29 <0.001 >3 cm 0.32 0.17 0.83 0.21 Histology ADC 0.72 0.25 0.706 SCC 0.75 0.31 LCC 0.66 0.32 SCLC 0.65 0.30 NSCLC NOS 0.85 0.15 Stage I-II 0.68 0.31 0.258 III-IV 0.77 0.23 Status Alive 0.70 0.26 0.005 Death 0.84 0.22 .sup.1Arbitrary units. .sup.2Mann-Whitney U test or Kruskal Wallis test.

[0075] The next aim was to generate diagnostic models based on both molecular markers and clinical features. First, univariate logistic regression analyses were performed for these last features (Table 16).

TABLE-US-00017 TABLE 16 Univariate logistic regression for the evaluation of clinical variables as markers for discrimination of pulmonary nodules. Marker LR chi.sup.2 p value Sex 0.87 0.352 Age 14.98 <0.001 Smoking status 17.57 <0.001 Pack-years 11.12 <0.001 FEV1% Predicted 2.42 0.120 Nodule size 27.18 <0.001

[0076] Based on the results obtained from these univariate analyses, clinical variables selected to be included in the model were age, smoking status, pack-years and nodule size. Two combined models were generated, one including the clinical variables and C4c, and another including the clinical variables and the three molecular markers. In the first case (C4c combined with clinical variables), the LR chi2 of the model was 69.81 (p<0.001). The area under the ROC curve was 0.88 (95% CI=0.83-0.94) (FIG. 9A). The LR chi2 of the model based on the three molecular markers and the clinical variables was 86.82 (p<0.001). The area under the ROC curve was 0.92 (95% CI=0.87-0.96) (FIG. 9B). Other diagnostic characteristics of the two models are shown in Table 17.

TABLE-US-00018 TABLE 17 Performance of lung cancer diagnostic models based on the plasma levels of molecular markers (C4c alone, or together with CYFRA 21-1 and CRP) and clinical variables (age, smoking status, pack-years and nodule size). Clinical Clinical variables + variables + C4c/CYFRA C4c 21-1/CRP Sensitivity 87% 91% Specificity 76% 76% Positive predictive value.sup.1 13% 14% Negative predictive value.sup.1 99.3% 99.5% Likelihood positive ratio 3.60 3.76 Likelihood negative ratio 0.17 0.12 Correctly classified 82% 84% .sup.1Positive and negative predictive values were calculated with an estimated prevalence of malignant nodules in this clinical setting of 4%.

[0077] Noninvasive diagnostic models for lung cancer based on clinical and image characteristics have been described. Gould et al. developed and validated a clinical model to discriminate lung cancer from benign lung nodules using age, smoking history, nodule diameter and time since quitting smoking. This model has been previously used to evaluate the diagnostic value added by molecular signatures to clinical and chest CT data for the noninvasive diagnosis of patients presenting indeterminate pulmonary nodules. Similarly, we evaluated the capacity of C4c, alone or in combination with CYFRA 21-1 and CRP, to complement this validated clinical model for the detection of lung cancer in patients with indeterminate pulmonary nodules.

[0078] We first assessed the diagnostic performance of the Gould's clinical model in our set of patients. Due to the limited clinical information available from some patients, the model could only be applied to 134 patients. A logistic regression analysis yielded an LR chi2 of 58.19 (p<0.001). The area under the ROC curve was 0.86 (95% CI=0.80-0.92) (FIG. 10A). Other diagnostic characteristics of the model are shown in Table 18.

TABLE-US-00019 TABLE 18 Performance of the Gould's model based on the clinical variables age, smoking history, nodule diameter and time since quitting smoking. Gould's model Sensitivity 86% Specificity 73% Positive predictive value.sup.1 12% Negative predictive value.sup.1 99.2% Likelihood positive ratio 3.15 Likelihood negative ratio 0.20 Correctly classified 80% .sup.1Positive and negative predictive values were calculated with an estimated prevalence of malignant nodules in this clinical setting of 4%.

[0079] The accuracy of the clinical model increased when C4c or C4c/CYFRA 21-1/CRP was added to the model. In the first case (Gould's model and C4c), the LR chi2 was 63.96 (p<0.001). The area under the ROC curve was 0.87 (95% CI=0.81-0.93) (FIG. 10B). In the second case (Gould's model and C4c/CYFRA/CRP), the LR chi2 was 86.91 (p<0.001). The area under the ROC curve was 0.92 (95% CI=0.87-0.96) (FIG. 10C). Other diagnostic characteristics of the two models are shown in Table 19.

TABLE-US-00020 TABLE 19 Performance of lung cancer diagnostic models based on the plasma levels of molecular markers (C4c alone, or together with CYFRA 21-1 and CRP) and variables from the clinically validated Gould's model (age, smoking history, nodule diameter and time since quitting smoking). Gould's Gould's model + model + C4c/CYFRA C4c 21-1/CRP Sensitivity 92% 92% Specificity 66% 72% Positive predictive value.sup.1 10% 12% Negative predictive value.sup.1 99.5% 99.6% Likelihood positive ratio 2.67 3.34 Likelihood negative ratio 0.12 0.11 Correctly classified 81% 84% .sup.1Positive and negative predictive values were calculated with an estimated prevalence of malignant nodules in this clinical setting of 4%.

Example 4. Screening of Individuals at High-Risk for Lung Cancer.

[0080] Material and Methods

[0081] 128 Asymptomatic smokers over the age of 40 were enrolled for the CT screening program I-ELCAP at the Clinica Universidad de Navarra (CUN). 32 were diagnosed with lung cancer in the context of the program, and the remaining 96 had no evidence of lung cancer after CT-screening program. Both groups are matched by sex, age, and smoking history. Plasma levels of C4c, C4c/CYFRA and C4c/CYFRA/CRP were measured in those high-risk patients finally diagnosed with lung cancer and compared with high-risk patients with no evidence of lung cancer.

[0082] Results

[0083] Plasma levels of C4c, C4c/CYFRA and C4c/CYFRA/CRP were significantly higher in individuals at high-risk for lung cancer which were finally diagnosed with lung cancer. FIG. 11 shows ROC curves obtained from the diagnostic model based on the quantification of C4c (A), C4c/CYFRA (B) and C4c/CYFRA/CRP (C) in plasma samples from the screening of individuals at high-risk for lung cancer.

[0084] Conclusion

[0085] These analyses evidence the diagnostic capacity of the quantification of C4c in plasma samples (alone or in combination with other molecular markers and epidemiological and clinical features) for discriminating which pulmonary nodules are malignant. The application of this multimodality approach predicts lung cancer more accurately and may be used for the identification of the subjects that need more active follow-up, which would improve the clinical management of lung nodules by reducing the number of unnecessary procedures.