PREDICTION OF 5-YEAR SURVIVAL OF NON-SMALL CELL LUNG CANCER PATIENTS WITH STAGE III AFTER COMPLETE RESECTIONS Oleg Kshivets, MD, PhD  Department of Surgery, Siauliai Cancer Center, Siauliai, Lithuania 2004 ASCO Annual Meeting, New Orleans, Louisiana, the USA, June 3-8, 2004
Abstract PREDICTION OF 5-YEAR SURVIVAL OF NON-SMALL CELL LUNG CANCER PATIENTS WITH STAGE III AFTER COMPLETE RESECTIONS  Oleg Kshivets Surgery Department, Siauliai Cancer Center, Lithuania OBJECTIVE:  The influence of clinicomorphological data on 5-year survival (5YS) and life span (LS) of non-small lung cancer (LC) patients (LCP) with stage IIIA and IIIB after radical resections (R0) was investigated. METHODS:  In trial (1985-2003) the data of consecutive 139 LCP (age=57.8±0.6 years; male=129, female=10; tumor diameter: D=5.6±0.2 cm; pneumonectomy=75, upper lobectomy=40, lower lobectomy=17, middle lobectomy=1, bilobectomy=6, combined procedures with resection of pericardium, left atrium, aorta, v. cava superior, carina, diaphragm, ribs=67; only surgery=54, adjuvant chemoimmunoradiotherapy-AT=40: CAV/gemzar-cisplatin + thymalin/taktivin + radiotherapy=40, postoperative radiotherapy-PR=45) with stage III (squamous cell=94, adenocarcinoma=34, large cell=11; stage IIIA=109, stage IIIB=30; T1=11, T2=48, T3=50, T4=30; N0=13, N1=32, N2=94; G1=24, G2=31, G3=84) was reviewed. Variables selected for 5YS study were input levels of blood, biochemic and hemostatic factors, sex, age, TNMG, D. Survival curves were estimated by Kaplan-Meier method. Differences in curves between groups were evaluated using a log-rank test. Neural networks computing, Cox regression, clustering, discriminant analysis, structural equation modeling and Monte Carlo simulation were used to determine any significant regularity. RESULTS:  General life span was 1019.3  82.9 days (median=519). 49 LCP (35.3%) were alive, 44 from them lived more than 5 years (31.7%) without any features of progressing. 87 LCP (62.6%) died because of relapses and generalization during the first 5 years after surgery. Cox modeling displayed that 5-year survival of LCP (n=139) after complete resections significantly depended on: N0-2 (P=0.027), histology (P=0.002), G1-3 (P=0.036), character of operation (P=0.010), AT (P=0.0003), 20 blood factors (P=0.001-0.044). Neural networks computing and genetic algorithm selection revealed relationships between 5-year survival of LCP and AT (rank=1), G1-3 (2), PR (3), N0-2 (4), tumor growth (5), histology (6), procedure type (7), stage IIIA/IIIB (8), percent of monocytes (9), ESS (10).  Correct prediction of LCP survival after radical procedures was 75.6% by logistic regression (odds ratio=7.57), 85.5% by discriminant analysis and 100% by neural networks computing (area under ROC curve=1.0; error=0.0027).
Factors: 1) Antropometric Factors…………... 4 2) Blood Analysis…………………... 26 3) Hemostasis Factors………………. 8 4) Cell Ratio Factors………………... 9  5) Lung Cancer Characteristics……. 8 6) Biochemic Factors………………... 5 7) Treatment Characteristics………. 5 8) Survival Data……………………... 3 In All……………………………….68
Main Problem of Analysis of Alive Supersystems (e.g. Lung Cancer Patient Homeostasis):  Phenomenon of «Combinatorial Explosion» Number of Factors:……………………………...…..  68 Number of Possible Combination for Random Search:……………..…………………..  n!=68!=2.5e+96   Operation Time of The 7G Superteracomputer (1000TFLOPS) (The 21 st  Century)……. 7.9e+73 Years Age of Universe………………………….1.3e+13 Years
Basis: NP     RP     P        n!   n*n*2(e+n)  or  n log n    n          AI     CSA+S+B     SM
Radical Procedures: Pneumonectomy……………….. 75 Upper/Lower Bilobectomy…...… 6 Upper Lobectomy…………...… 40 Lower Lobectomy……………... 17 Middle Lobectomy………….…... 1 In All…………………………...139
Combined &   E xtensive  R adical  P rocedures with  R esection of  P ericardium,  L eft  A trium,  A orta,  V ena  C ava  S uperior,  V ena  A zygos,  C arina,  Trachea, D iaphragm,  C hest  W all , Ribs, etc…..……………………. 67 Sistematic Mediastinal Lymph Node-N2 Dissection………….. 139
Staging: T1…….11  N0..…..13  G1…..24 T2…….48  N1……32   G2…..31 T3…….50  N2……94   G3…..84 T4…….30  Stage IIIA...109  Stage IIIB...30 Squamous Cell Carcinoma…..………….94 Adenocarcinoma…………………………34 Large Cell Carcinoma…………………...11
Samplings: Adjuvant Chemoimmunoradiotherapy …….. 40 P/O Radiotherapy …...……………………….. 45 Surgery Alone ……..………………………….. 54 In All…………………………………………..139 Male……………………………………….…..129  Female……...…………………………………..10 Age=57.8±0.6 years
Adjuvant Therapy after Complete Resections Adjuvant Chemoimmunoradiotherapy:  1 cycle of bolus chemotherapy (CAVT) was initiated 10-14 days after resections and consisted of Cyclophosphamid 500 mg/m 2  IV on day 1, Doxorubicin 50 mg/m 2  IV on day 1, Vincristin 1.4 mg/m 2  IV on day 1. Immunotherapy consisted Thymalin or Taktivin 20 mg IM on days 1, 2, 3, 4 and 5. Chest radiotherapy (45-50 Gy) was administered since 7 day after 1 cycle chemoimmunotherapy at a daily dose of 1.8-2 Gy. No prophylactic cranial irradiation was used. From 2 to 3 weeks after completion of radiotherapy 3-4 courses of CAVT were repeated every 21-28 day. Chemotherapy by gemzar 1250 mg/m 2  IV on day 1, 8, 15 and cisplatin 75 mg/m 2  on day 1 was initiated on 14 day after surgery and was repeated every 14 day (5-6 courses).   P/o Radiotherapy:   Radiotherapy ( 60 CO; ROKUS, Russia) with a total tumor dose 45-50 Gy (2-4 weeks after surgery) consisted of single daily fractions of 180-200 cGy 5 days weekly. The treatment volume included the ipsilateral hilus, the supraclavicular fossa and the mediastinum from the incisura jugularis to 5-7 cm below the carina. The lower mediastinum was included in cases of primary tumors in the lower lobes. The resected tumor bed was included in all patients. Parallel-opposed AP-PA fields were used. All fields were checked using the treatment planning program COSPO. Doses were specified at middepth for parallel-opposed technique or at the intersection of central axes for oblique technique. No prophylactic cranial irradiation was used.
Survival Rate of Lung Cancer Patients with  Stage III after Complete Resections (n=139): Alive…………………………………………. 49 (35.3%) 5-Year Survivors……………………………. 44 (31.7%)   Losses from Lung Cancer………………….. 87 (62.6%) Life Span=1019.3±82.9 days  5-Year Survivors with Stage IIIA…………. 38 (34.9%) 5-Year Survivors with Stage IIIB…………... 6 (20%) 5-Year Survivors after Surgery alone……... 11(20.4%) 5-Year Survivors after P/o Radiotherapy.… 14(31.1%) 5-Year Survivors after Adjuvant Chemoimmunoradiotherapy……………….. 18 (45%) At All………….………………………………139 (100%)
Significant  Factors  between Lung Cancer Losses & 5-Year Survivors with Stage III (n=131) Factors Mean±SD Mean±SD (Survivors)  (Losses)  P n=44  n=87 Weight (kg) 72.3±12.3 66.8±12.2 0.016 Seg.Neutrophiles (%) 64.9±11.9 69.2±9.7 0.030 Monocytes (%) 5.5±2.8 4.3±2.8 0.021 Prothrombin Index (%) 94.8±6.9 99.0±8.2 0.004 Sul. Probe 1.95±0.17 1.86±0.19 0.009 Fibrinogen-B 1.16±0.37 1.44±0.87 0.046 Life Span (days) 2283.5±684.6 424.9±309.1 0.0000 Log-Rank Test  P N1 vs. N2 0.006 Ad.CHIRT 0.0005 Surgery Alone 0.015
Product-Limit (Kaplan-Maier) Analysis Results in Prediction of Lung Cancer Patients Survival with Stage III after Complete Resections (n=139)  Graph of Survival Times vs. Cum. Proportion Surviving
Cumulative Proportion Lung Cancer Patients Surviving (Kaplan-Meier) (n=139)
Cumulative Proportion Lung Cancer Patients Surviving (Kaplan-Meier)
Results of Multivariate Proportional Hazard Cox Regression Analysis:
Results of Multifactor Analysis in Prediction of Lung Cancer Patients Survival with Stage IIIA & IIIB (n=139)
Results of Discriminant Analysis in Prediction of Lung Cancer Patients Survival with Stage III after Complete Resections (n=131) Discriminant Function Analysis Summary   Wilks' Lambda: .57124  approx. F (17,113)=4.9890  p< 0.0000 Wilks'  Partial  F-remove   Lambda  Lambda  (1,113)    P-level  Ad. CHTITR .651484 .876837 15.87236  .000120 PI .597889 .955437 5.27047 .023535 N 0-2   .615918 .927469 8.83695 .003608 M .597130 .956651 5.12045 .025556 SUL .579428 .985878 1.61866 .205891 BILIR UBIN .603061 .947242 6.29376 .013535 FIBR - B .599580 .952742 5.60500 .019609 E / CC .583896 .978333 2.50258 .116455 HIST OLOGY .578991 .986621 1.53229 .218335 HEP .TOL .592369 .964340 4.17857 .043263 REC .T. .577855 .988561 1.30758 .255250 S tage IIIAB .591518 .965728 4.01021 .047620 Tumor Size .583061 .979735 2.33731 .129102 P/o RT  .579747 .985336 1.68174 .197335 TIM .580790 .983565 1.88818 .172127 GRO W TH .580624 .983847 1.85522 .175887 LYM .578408 .987616 1.41696 .236398
Results of Logistic Regression Analysis in Prediction of Lung Cancer Patients Survival with Stage III  after Complete Resections (n=131)   B  S.E. Wald  df Sig. Exp(B) 95.0% C.I.for EXP(B)  P Lower Upper GROWTH  -1.007  .600 2.813  1 .094 .365 .113 1.185 N0-2  10.212  2 .006 N0-2(1)   5.122  1.933 7.022  1 .008 167.660 3.795 7407.571 N0-2(2)   1.824  .716 6.482  1 .011 6.196 1.522 25.226 E   .223  .130 2.951  1 .086 1.250 .969 1.613 S   .139  .066 4.471  1 .034 1.149 1.010 1.306 M   .217  .097 4.999  1 .025 1.243 1.027 1.503 SABS   226.039  121.7 3.450  1 .063 1. 6 E+98 .000 5. 8E +201 COAG_B  -3.767  2.029 3.449  1 .063 .023 .000 1.232 PI   -.065  .038 2.935  1 .087 .937 .869 1.009 BILIRUBIN  -.550  .197 7.815  1 .005 .577 .392 .848 TIM   -1.019  .379 7.240  1 .007 .361 .172 .758 FIBR-B  -2.152  .896 5.773  1 .016 .116 .020 .673 HEP.TOL.   -.007  .003 7.733  1 .005 .993 .987 .998 AD.CHIRT   -2.342  .717 10.671  1 .001 .096 .024 .392 L/CC   1.664  .657 6.412  1 .011 5.279 1.456 19.137 S/CC   -2.664  1.062 6.291  1 .012 .070 .009 .559 STIIIAB  -3.027  1.612 3.526  1 .060 .048 .002 1.142 Constant  8.398  5.561 2.280  1 .131 4436.565
Results of Clustering in Prediction of Lung Cancer Patients  5-Year Survival (n=131)
Results of Clustering in Prediction of Lung Cancer Patients Survival  with Stage III after Complete Resections (n=131) Logic al  Formulas  based on Simple Mean 5YS. Losses :   44.00 <= Weight  (5.3%) <= 123.00  & 1.00 <= M  (4.8%) <= 13.00  &   74.00 <= PI  (7.6%) <= 118.00  & 1.30 <= Sul  (6.1%) <= 2.20  & not CHIRT.Grad 2(6.7%) Objects 87  Error1 = 0.16 (14)  Error2 = 0.55 (24) 5YS. Survivors: 46.00 <= Weight  (5.3%) <= 103.00 & 1.00 <= M  (4.8%) <= 17.00  & 82.00 <= PI  (7.6%) <= 110.00  & 1.60 <= Sul  (6.1%) <= 2.30  & CHIRT.Grad 2(6.7%) Objects 44  Error1 = 0.59 (26)  Error2 = 0.15 (13) Logic Formulas  based on M edian 5YS. Losses:  44.00 <= Weight  (6.8%) <= 123.00  & 0.00 <= E  (10.8%) <= 10.00  & 3.00 <= Glu  (6.0%) <= 7.60  & 1.30 <= Sul  (6.1%) <= 2.20  & 0.03 <= M/cc  (7.3%) <= 2.90 Objects 87  Error1 = 0.00 (0)  Error2 = 0.93 (41) 5YS. Survivors:   Growth.Grad 1(11.9%)  & also not Growth.Grad 2(11.1%)  & 39.00 <= S  (4.7%) <= 83.00  & 1.00 <= M  (8.1%) <= 17.00  & 82.00 <= PI  (10.9%) <= 110.00 Objects 44  Error1 = 0.39 (17)  Error2 = 0.37 (32) Logic Formulas  based on Semi-Range 5YS. Losses 170.00 <= Thr  (10.7%) <= 681.00  & 1.00 <= M  (25.6%) <= 13.00  & 60.00 <= Rec.T.  (27.3%) <= 246.00  Objects 87  Error1 = 0.00 (0)  Error2 = 0.86 (38) 5YS. Survivors 3.10 <= Er  (3.8%) <= 5.10  & 0.80 <= CI  (4.0%) <= 1.00  & 0.00 <= P/cc  (3.8%) <= 0.44  & 9.98 <= Ertot  (3.8%) <= 36.77  & 0.00 <= Ptot  (3.8%) <= 2.96 Objects 44  Error1 = 0.00 (0)  Error2 = 0.86 (75)
Results of  Correspondence  Analysis in Prediction of  Lung  Cancer Patients Survival with Stage III (n=131)
Results of Bootstrap Resampling in Prediction of 5-Year Survival of Lung Cancer Patients with Stage III (n=131; subsamplings=935)
Results of Bootstrap Resampling in Prediction of 5-Year Survival of Lung Cancer Patients with Stage III (n=131; subsamplings=935)
Results of  Bootstrap   Simulation  in Prediction of Lung Cancer Patients Survival   with Stage III after Complete Resections (n=131 ; subsamplings=935 )   Factor Rank Kendall’s   P Tau-A  PI 1 - 0. 1479 0 . 00000 Mtot 2 0. 1474 0 . 00000 Ertot 3 0. 1423 0 . 00000 Weight 4 0. 1363 0 . 00000 M% 5 0.1 344 0 . 00000 M/CC 6 0.1 280 0 .0 0000 Etot 7 0.12 0 1 0 . 00000 Mabs 8 0.11 78 0 . 00000 Sul 9 0. 1127 0 . 00000 Ad.CHIRT 10 0. 1116 0 . 00000 Surgery Alone 11 - 0. 1 0 98 0 . 00000 Eabs 12 0.0 997 0 .0 0001 N 13 - 0.0 986 0 .0 0001 E % 14 0.0 978 0 . 00001 S% 15 - 0.0 955 0.00 001 Factor Rank Kendall’s   P Tau-A   E/CC 16 0.0 951 0 . 00002 Glucose 17 0.0 937 0 .0 0002 Er 18 0.0 9 23 0 . 00003 Growth 19 - 0.0 898 0 . 00003 Thrtot 20 0.0 814 0 .0 004 Tim 21 - 0.0 770 0 .0 005 Bilirubim 22 - 0.0 763 0 .0 005 Lymtot 23 0.07 59 0 .0 005 T 24 0.06 69 0 . 00 2 ESS 25 - 0.0 61 5 0 . 005 Hep.Tol. 26 - 0.05 77 0 . 0 1 1 G 27 - 0.0 545 0 . 013 Fibrinogen-B 28 - 0.0 539 0 .0 15 D 29 0.0 536 0 . 016 Lym% 30 0.0 5 0 0 0 . 0 30 Operation 31 -0.0445 0.048
Results of Neural Networks Computing in Prediction of Lung Cancer Patients Survival   with Stage III after Complete Resections (n=131)   Factor Rank Error Ratio Ad.CHIRT 1 0.297 107.88 G 2 0.242 88.091 P/o RT 3 0.212 76.997 N 4 0.196 71.221 Growth 5 0.171 62.357 Histology 6 0.157 57.028 Operation 7 0.121 43.899 Stage IIIAB 8 0.113 41.246 M(%) 9 0.089 32.447 ESS 10 0.074 26.929 PI 11 0.070 25.534 T 12 0.058 21.062 HC/CC 13 0.052 19.032 Eabs 14 0.034 12.239 Hep.Tol. 15 0.028 10.040 Factor Rank Error Ratio S(%) 16 0.027 9.849 E/CC 17 0.025 9.066 Surgery Alone 18 0.023 8.537 Sex 19 0.010 3.731 Lym(%) 20 0.009 3.120 Fibrinogen 21 0.008 3.074 E(%) 22 0.008 3.046 Glucose 23 0.007 2.490 Fibrinogen-B 24 0.006 2.201 Age 25 0.005 1.756 Thr/CC 26 0.005 1.671 Hb 27 0.004 1.447 M/CC 28 0.004 1.400 Hem.T. 29 0.004 1.313 Weight 30 0.004 1.305
Neural Networks in Prediction of Lung Cancer Patients Survival (n=131) Losses   5-year survivors  Baseline Errors=0.000127; Total   87   44   Area under ROC curve=1 .000;  Correct   87   44   Correct Classification Rate=100% Wrong  0   0 Unknown   0   0 Losses   87    0 Alive   0   44 Genetic Algorithm Selection Useful for   Gr  N  G  Hb  E  S  Lym  M  ESS Glu PI  T/CC E/CC M/CC Surg  AT Survival   YesYes Yes Yes Yes Yes  Yes Yes Yes  Yes Yes Yes  Yes  Yes  Yes  Yes
Prediction of  Lung  Cancer Patients Survival with Stage III after Complete Resections (n=131) Classification of Cases by Logistic Regression, n=131 (5-Year Survivors--Losses) Observed  Pred.Losses  Pred.Survivors  Correct Losses   76  11  87.4% 5-Year Survivors  21  23  52.3% Total  97  34  75 . 6 % Classification of Cases by  General  Discriminant Analysis, n=131 (5-Year Survivors--Losses) Observed  Pred.Losses  Pred.Survivors  Correct Losses   79  8   90.8% 5-Year Survivors  11  33  75.0% Total  90  41  85 .5%
Prediction of  Lung  Cancer Patients Survival with Stage III after Complete Resections (n=131) Classification of Cases by Clastering, n=131 (5-Year Survivors--Losses) Observed  Pred.Losses  Pred.Survivors  Correct Losses   81  6  93.1% 5-Year Survivors  9  35  79.5% Total  90  41  88 .5% Classification of Cases by Neural Networks, n=131 (5-Year Survivors--Losses) Observed  Pred.Losses  Pred.Survivors  Correct Losses   87  0   100.0% 5-Year Survivors  0  44  100.0% Total  87  44  100 .0%
Lymphocyte  &  Monocyte  Circuit in  Prediction  of Lung Cancer  Patients Survival  with Stage III (n=131)
Ratio of Segmented Neutrophile and Cancer Cell Populations  &  Blood Glucose Level  in  Prediction  of Lung Cancer  Patients Survival with Stage III  (n=131)
Weight  &  Prothrombin Index  in  Prediction  of Lung Cancer  Patients Survival with Stage III  (n=131)
Significant Networks between Clinicopathologic, Biochemic, Hemostasis & Hematologic Data and 5-Year Survival of Lung cancer Patients with Stage III (n=131)
Results of Monte Carlo Simulation in Prediction of Lung Cancer Patients Survival with Stage III (n=131)
SEPATH Networks in Prediction of Lung Cancer Patients Survival with Stage III (n=131)
Holling-Tenner Models of Alive Supersystem “Lung Cancer-Cytotoxic Cell Population ”
Lung Cancer Dynamics
 
Conclusions: It was revealed that 5-year survival of lung cancer patients with stage III after complete pulmonary resections significantly depended on:  1) lung cancer characteristics; 2) adjuvant treatment;  3) level of blood cell circuit;  4) cell ratio factors;  5) hemostasis system;  6) biochemic homeostasis.
Address: Oleg Kshivets, M.D., Ph.D. Thoracic Surgeon Department of Surgery Siauliai Cancer Center Tilzes:42-16, 5400 Siauliai, Lithuania Tel. (37041)416614 kshivets@yahoo.com  http//:myprofile.cos.com/Kshivets

Kshivets O. Lung Cancer Stage III Surgery

  • 1.
    PREDICTION OF 5-YEARSURVIVAL OF NON-SMALL CELL LUNG CANCER PATIENTS WITH STAGE III AFTER COMPLETE RESECTIONS Oleg Kshivets, MD, PhD Department of Surgery, Siauliai Cancer Center, Siauliai, Lithuania 2004 ASCO Annual Meeting, New Orleans, Louisiana, the USA, June 3-8, 2004
  • 2.
    Abstract PREDICTION OF5-YEAR SURVIVAL OF NON-SMALL CELL LUNG CANCER PATIENTS WITH STAGE III AFTER COMPLETE RESECTIONS Oleg Kshivets Surgery Department, Siauliai Cancer Center, Lithuania OBJECTIVE: The influence of clinicomorphological data on 5-year survival (5YS) and life span (LS) of non-small lung cancer (LC) patients (LCP) with stage IIIA and IIIB after radical resections (R0) was investigated. METHODS: In trial (1985-2003) the data of consecutive 139 LCP (age=57.8±0.6 years; male=129, female=10; tumor diameter: D=5.6±0.2 cm; pneumonectomy=75, upper lobectomy=40, lower lobectomy=17, middle lobectomy=1, bilobectomy=6, combined procedures with resection of pericardium, left atrium, aorta, v. cava superior, carina, diaphragm, ribs=67; only surgery=54, adjuvant chemoimmunoradiotherapy-AT=40: CAV/gemzar-cisplatin + thymalin/taktivin + radiotherapy=40, postoperative radiotherapy-PR=45) with stage III (squamous cell=94, adenocarcinoma=34, large cell=11; stage IIIA=109, stage IIIB=30; T1=11, T2=48, T3=50, T4=30; N0=13, N1=32, N2=94; G1=24, G2=31, G3=84) was reviewed. Variables selected for 5YS study were input levels of blood, biochemic and hemostatic factors, sex, age, TNMG, D. Survival curves were estimated by Kaplan-Meier method. Differences in curves between groups were evaluated using a log-rank test. Neural networks computing, Cox regression, clustering, discriminant analysis, structural equation modeling and Monte Carlo simulation were used to determine any significant regularity. RESULTS: General life span was 1019.3  82.9 days (median=519). 49 LCP (35.3%) were alive, 44 from them lived more than 5 years (31.7%) without any features of progressing. 87 LCP (62.6%) died because of relapses and generalization during the first 5 years after surgery. Cox modeling displayed that 5-year survival of LCP (n=139) after complete resections significantly depended on: N0-2 (P=0.027), histology (P=0.002), G1-3 (P=0.036), character of operation (P=0.010), AT (P=0.0003), 20 blood factors (P=0.001-0.044). Neural networks computing and genetic algorithm selection revealed relationships between 5-year survival of LCP and AT (rank=1), G1-3 (2), PR (3), N0-2 (4), tumor growth (5), histology (6), procedure type (7), stage IIIA/IIIB (8), percent of monocytes (9), ESS (10). Correct prediction of LCP survival after radical procedures was 75.6% by logistic regression (odds ratio=7.57), 85.5% by discriminant analysis and 100% by neural networks computing (area under ROC curve=1.0; error=0.0027).
  • 3.
    Factors: 1) AntropometricFactors…………... 4 2) Blood Analysis…………………... 26 3) Hemostasis Factors………………. 8 4) Cell Ratio Factors………………... 9 5) Lung Cancer Characteristics……. 8 6) Biochemic Factors………………... 5 7) Treatment Characteristics………. 5 8) Survival Data……………………... 3 In All……………………………….68
  • 4.
    Main Problem ofAnalysis of Alive Supersystems (e.g. Lung Cancer Patient Homeostasis): Phenomenon of «Combinatorial Explosion» Number of Factors:……………………………...….. 68 Number of Possible Combination for Random Search:……………..………………….. n!=68!=2.5e+96 Operation Time of The 7G Superteracomputer (1000TFLOPS) (The 21 st Century)……. 7.9e+73 Years Age of Universe………………………….1.3e+13 Years
  • 5.
    Basis: NP  RP  P    n!  n*n*2(e+n) or n log n  n    AI  CSA+S+B  SM
  • 6.
    Radical Procedures: Pneumonectomy………………..75 Upper/Lower Bilobectomy…...… 6 Upper Lobectomy…………...… 40 Lower Lobectomy……………... 17 Middle Lobectomy………….…... 1 In All…………………………...139
  • 7.
    Combined & E xtensive R adical P rocedures with R esection of P ericardium, L eft A trium, A orta, V ena C ava S uperior, V ena A zygos, C arina, Trachea, D iaphragm, C hest W all , Ribs, etc…..……………………. 67 Sistematic Mediastinal Lymph Node-N2 Dissection………….. 139
  • 8.
    Staging: T1…….11 N0..…..13 G1…..24 T2…….48 N1……32 G2…..31 T3…….50 N2……94 G3…..84 T4…….30 Stage IIIA...109 Stage IIIB...30 Squamous Cell Carcinoma…..………….94 Adenocarcinoma…………………………34 Large Cell Carcinoma…………………...11
  • 9.
    Samplings: Adjuvant Chemoimmunoradiotherapy…….. 40 P/O Radiotherapy …...……………………….. 45 Surgery Alone ……..………………………….. 54 In All…………………………………………..139 Male……………………………………….…..129 Female……...…………………………………..10 Age=57.8±0.6 years
  • 10.
    Adjuvant Therapy afterComplete Resections Adjuvant Chemoimmunoradiotherapy: 1 cycle of bolus chemotherapy (CAVT) was initiated 10-14 days after resections and consisted of Cyclophosphamid 500 mg/m 2 IV on day 1, Doxorubicin 50 mg/m 2 IV on day 1, Vincristin 1.4 mg/m 2 IV on day 1. Immunotherapy consisted Thymalin or Taktivin 20 mg IM on days 1, 2, 3, 4 and 5. Chest radiotherapy (45-50 Gy) was administered since 7 day after 1 cycle chemoimmunotherapy at a daily dose of 1.8-2 Gy. No prophylactic cranial irradiation was used. From 2 to 3 weeks after completion of radiotherapy 3-4 courses of CAVT were repeated every 21-28 day. Chemotherapy by gemzar 1250 mg/m 2 IV on day 1, 8, 15 and cisplatin 75 mg/m 2 on day 1 was initiated on 14 day after surgery and was repeated every 14 day (5-6 courses). P/o Radiotherapy: Radiotherapy ( 60 CO; ROKUS, Russia) with a total tumor dose 45-50 Gy (2-4 weeks after surgery) consisted of single daily fractions of 180-200 cGy 5 days weekly. The treatment volume included the ipsilateral hilus, the supraclavicular fossa and the mediastinum from the incisura jugularis to 5-7 cm below the carina. The lower mediastinum was included in cases of primary tumors in the lower lobes. The resected tumor bed was included in all patients. Parallel-opposed AP-PA fields were used. All fields were checked using the treatment planning program COSPO. Doses were specified at middepth for parallel-opposed technique or at the intersection of central axes for oblique technique. No prophylactic cranial irradiation was used.
  • 11.
    Survival Rate ofLung Cancer Patients with Stage III after Complete Resections (n=139): Alive…………………………………………. 49 (35.3%) 5-Year Survivors……………………………. 44 (31.7%) Losses from Lung Cancer………………….. 87 (62.6%) Life Span=1019.3±82.9 days 5-Year Survivors with Stage IIIA…………. 38 (34.9%) 5-Year Survivors with Stage IIIB…………... 6 (20%) 5-Year Survivors after Surgery alone……... 11(20.4%) 5-Year Survivors after P/o Radiotherapy.… 14(31.1%) 5-Year Survivors after Adjuvant Chemoimmunoradiotherapy……………….. 18 (45%) At All………….………………………………139 (100%)
  • 12.
    Significant Factors between Lung Cancer Losses & 5-Year Survivors with Stage III (n=131) Factors Mean±SD Mean±SD (Survivors) (Losses) P n=44 n=87 Weight (kg) 72.3±12.3 66.8±12.2 0.016 Seg.Neutrophiles (%) 64.9±11.9 69.2±9.7 0.030 Monocytes (%) 5.5±2.8 4.3±2.8 0.021 Prothrombin Index (%) 94.8±6.9 99.0±8.2 0.004 Sul. Probe 1.95±0.17 1.86±0.19 0.009 Fibrinogen-B 1.16±0.37 1.44±0.87 0.046 Life Span (days) 2283.5±684.6 424.9±309.1 0.0000 Log-Rank Test P N1 vs. N2 0.006 Ad.CHIRT 0.0005 Surgery Alone 0.015
  • 13.
    Product-Limit (Kaplan-Maier) AnalysisResults in Prediction of Lung Cancer Patients Survival with Stage III after Complete Resections (n=139) Graph of Survival Times vs. Cum. Proportion Surviving
  • 14.
    Cumulative Proportion LungCancer Patients Surviving (Kaplan-Meier) (n=139)
  • 15.
    Cumulative Proportion LungCancer Patients Surviving (Kaplan-Meier)
  • 16.
    Results of MultivariateProportional Hazard Cox Regression Analysis:
  • 17.
    Results of MultifactorAnalysis in Prediction of Lung Cancer Patients Survival with Stage IIIA & IIIB (n=139)
  • 18.
    Results of DiscriminantAnalysis in Prediction of Lung Cancer Patients Survival with Stage III after Complete Resections (n=131) Discriminant Function Analysis Summary Wilks' Lambda: .57124 approx. F (17,113)=4.9890 p< 0.0000 Wilks' Partial F-remove Lambda Lambda (1,113) P-level Ad. CHTITR .651484 .876837 15.87236 .000120 PI .597889 .955437 5.27047 .023535 N 0-2 .615918 .927469 8.83695 .003608 M .597130 .956651 5.12045 .025556 SUL .579428 .985878 1.61866 .205891 BILIR UBIN .603061 .947242 6.29376 .013535 FIBR - B .599580 .952742 5.60500 .019609 E / CC .583896 .978333 2.50258 .116455 HIST OLOGY .578991 .986621 1.53229 .218335 HEP .TOL .592369 .964340 4.17857 .043263 REC .T. .577855 .988561 1.30758 .255250 S tage IIIAB .591518 .965728 4.01021 .047620 Tumor Size .583061 .979735 2.33731 .129102 P/o RT .579747 .985336 1.68174 .197335 TIM .580790 .983565 1.88818 .172127 GRO W TH .580624 .983847 1.85522 .175887 LYM .578408 .987616 1.41696 .236398
  • 19.
    Results of LogisticRegression Analysis in Prediction of Lung Cancer Patients Survival with Stage III after Complete Resections (n=131) B S.E. Wald df Sig. Exp(B) 95.0% C.I.for EXP(B) P Lower Upper GROWTH -1.007 .600 2.813 1 .094 .365 .113 1.185 N0-2 10.212 2 .006 N0-2(1) 5.122 1.933 7.022 1 .008 167.660 3.795 7407.571 N0-2(2) 1.824 .716 6.482 1 .011 6.196 1.522 25.226 E .223 .130 2.951 1 .086 1.250 .969 1.613 S .139 .066 4.471 1 .034 1.149 1.010 1.306 M .217 .097 4.999 1 .025 1.243 1.027 1.503 SABS 226.039 121.7 3.450 1 .063 1. 6 E+98 .000 5. 8E +201 COAG_B -3.767 2.029 3.449 1 .063 .023 .000 1.232 PI -.065 .038 2.935 1 .087 .937 .869 1.009 BILIRUBIN -.550 .197 7.815 1 .005 .577 .392 .848 TIM -1.019 .379 7.240 1 .007 .361 .172 .758 FIBR-B -2.152 .896 5.773 1 .016 .116 .020 .673 HEP.TOL. -.007 .003 7.733 1 .005 .993 .987 .998 AD.CHIRT -2.342 .717 10.671 1 .001 .096 .024 .392 L/CC 1.664 .657 6.412 1 .011 5.279 1.456 19.137 S/CC -2.664 1.062 6.291 1 .012 .070 .009 .559 STIIIAB -3.027 1.612 3.526 1 .060 .048 .002 1.142 Constant 8.398 5.561 2.280 1 .131 4436.565
  • 20.
    Results of Clusteringin Prediction of Lung Cancer Patients 5-Year Survival (n=131)
  • 21.
    Results of Clusteringin Prediction of Lung Cancer Patients Survival with Stage III after Complete Resections (n=131) Logic al Formulas based on Simple Mean 5YS. Losses : 44.00 <= Weight (5.3%) <= 123.00 & 1.00 <= M (4.8%) <= 13.00 & 74.00 <= PI (7.6%) <= 118.00 & 1.30 <= Sul (6.1%) <= 2.20 & not CHIRT.Grad 2(6.7%) Objects 87 Error1 = 0.16 (14) Error2 = 0.55 (24) 5YS. Survivors: 46.00 <= Weight (5.3%) <= 103.00 & 1.00 <= M (4.8%) <= 17.00 & 82.00 <= PI (7.6%) <= 110.00 & 1.60 <= Sul (6.1%) <= 2.30 & CHIRT.Grad 2(6.7%) Objects 44 Error1 = 0.59 (26) Error2 = 0.15 (13) Logic Formulas based on M edian 5YS. Losses: 44.00 <= Weight (6.8%) <= 123.00 & 0.00 <= E (10.8%) <= 10.00 & 3.00 <= Glu (6.0%) <= 7.60 & 1.30 <= Sul (6.1%) <= 2.20 & 0.03 <= M/cc (7.3%) <= 2.90 Objects 87 Error1 = 0.00 (0) Error2 = 0.93 (41) 5YS. Survivors: Growth.Grad 1(11.9%) & also not Growth.Grad 2(11.1%) & 39.00 <= S (4.7%) <= 83.00 & 1.00 <= M (8.1%) <= 17.00 & 82.00 <= PI (10.9%) <= 110.00 Objects 44 Error1 = 0.39 (17) Error2 = 0.37 (32) Logic Formulas based on Semi-Range 5YS. Losses 170.00 <= Thr (10.7%) <= 681.00 & 1.00 <= M (25.6%) <= 13.00 & 60.00 <= Rec.T. (27.3%) <= 246.00 Objects 87 Error1 = 0.00 (0) Error2 = 0.86 (38) 5YS. Survivors 3.10 <= Er (3.8%) <= 5.10 & 0.80 <= CI (4.0%) <= 1.00 & 0.00 <= P/cc (3.8%) <= 0.44 & 9.98 <= Ertot (3.8%) <= 36.77 & 0.00 <= Ptot (3.8%) <= 2.96 Objects 44 Error1 = 0.00 (0) Error2 = 0.86 (75)
  • 22.
    Results of Correspondence Analysis in Prediction of Lung Cancer Patients Survival with Stage III (n=131)
  • 23.
    Results of BootstrapResampling in Prediction of 5-Year Survival of Lung Cancer Patients with Stage III (n=131; subsamplings=935)
  • 24.
    Results of BootstrapResampling in Prediction of 5-Year Survival of Lung Cancer Patients with Stage III (n=131; subsamplings=935)
  • 25.
    Results of Bootstrap Simulation in Prediction of Lung Cancer Patients Survival with Stage III after Complete Resections (n=131 ; subsamplings=935 ) Factor Rank Kendall’s P Tau-A PI 1 - 0. 1479 0 . 00000 Mtot 2 0. 1474 0 . 00000 Ertot 3 0. 1423 0 . 00000 Weight 4 0. 1363 0 . 00000 M% 5 0.1 344 0 . 00000 M/CC 6 0.1 280 0 .0 0000 Etot 7 0.12 0 1 0 . 00000 Mabs 8 0.11 78 0 . 00000 Sul 9 0. 1127 0 . 00000 Ad.CHIRT 10 0. 1116 0 . 00000 Surgery Alone 11 - 0. 1 0 98 0 . 00000 Eabs 12 0.0 997 0 .0 0001 N 13 - 0.0 986 0 .0 0001 E % 14 0.0 978 0 . 00001 S% 15 - 0.0 955 0.00 001 Factor Rank Kendall’s P Tau-A E/CC 16 0.0 951 0 . 00002 Glucose 17 0.0 937 0 .0 0002 Er 18 0.0 9 23 0 . 00003 Growth 19 - 0.0 898 0 . 00003 Thrtot 20 0.0 814 0 .0 004 Tim 21 - 0.0 770 0 .0 005 Bilirubim 22 - 0.0 763 0 .0 005 Lymtot 23 0.07 59 0 .0 005 T 24 0.06 69 0 . 00 2 ESS 25 - 0.0 61 5 0 . 005 Hep.Tol. 26 - 0.05 77 0 . 0 1 1 G 27 - 0.0 545 0 . 013 Fibrinogen-B 28 - 0.0 539 0 .0 15 D 29 0.0 536 0 . 016 Lym% 30 0.0 5 0 0 0 . 0 30 Operation 31 -0.0445 0.048
  • 26.
    Results of NeuralNetworks Computing in Prediction of Lung Cancer Patients Survival with Stage III after Complete Resections (n=131) Factor Rank Error Ratio Ad.CHIRT 1 0.297 107.88 G 2 0.242 88.091 P/o RT 3 0.212 76.997 N 4 0.196 71.221 Growth 5 0.171 62.357 Histology 6 0.157 57.028 Operation 7 0.121 43.899 Stage IIIAB 8 0.113 41.246 M(%) 9 0.089 32.447 ESS 10 0.074 26.929 PI 11 0.070 25.534 T 12 0.058 21.062 HC/CC 13 0.052 19.032 Eabs 14 0.034 12.239 Hep.Tol. 15 0.028 10.040 Factor Rank Error Ratio S(%) 16 0.027 9.849 E/CC 17 0.025 9.066 Surgery Alone 18 0.023 8.537 Sex 19 0.010 3.731 Lym(%) 20 0.009 3.120 Fibrinogen 21 0.008 3.074 E(%) 22 0.008 3.046 Glucose 23 0.007 2.490 Fibrinogen-B 24 0.006 2.201 Age 25 0.005 1.756 Thr/CC 26 0.005 1.671 Hb 27 0.004 1.447 M/CC 28 0.004 1.400 Hem.T. 29 0.004 1.313 Weight 30 0.004 1.305
  • 27.
    Neural Networks inPrediction of Lung Cancer Patients Survival (n=131) Losses 5-year survivors Baseline Errors=0.000127; Total 87 44 Area under ROC curve=1 .000; Correct 87 44 Correct Classification Rate=100% Wrong 0 0 Unknown 0 0 Losses 87 0 Alive 0 44 Genetic Algorithm Selection Useful for Gr N G Hb E S Lym M ESS Glu PI T/CC E/CC M/CC Surg AT Survival YesYes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
  • 28.
    Prediction of Lung Cancer Patients Survival with Stage III after Complete Resections (n=131) Classification of Cases by Logistic Regression, n=131 (5-Year Survivors--Losses) Observed Pred.Losses Pred.Survivors Correct Losses 76 11 87.4% 5-Year Survivors 21 23 52.3% Total 97 34 75 . 6 % Classification of Cases by General Discriminant Analysis, n=131 (5-Year Survivors--Losses) Observed Pred.Losses Pred.Survivors Correct Losses 79 8 90.8% 5-Year Survivors 11 33 75.0% Total 90 41 85 .5%
  • 29.
    Prediction of Lung Cancer Patients Survival with Stage III after Complete Resections (n=131) Classification of Cases by Clastering, n=131 (5-Year Survivors--Losses) Observed Pred.Losses Pred.Survivors Correct Losses 81 6 93.1% 5-Year Survivors 9 35 79.5% Total 90 41 88 .5% Classification of Cases by Neural Networks, n=131 (5-Year Survivors--Losses) Observed Pred.Losses Pred.Survivors Correct Losses 87 0 100.0% 5-Year Survivors 0 44 100.0% Total 87 44 100 .0%
  • 30.
    Lymphocyte & Monocyte Circuit in Prediction of Lung Cancer Patients Survival with Stage III (n=131)
  • 31.
    Ratio of SegmentedNeutrophile and Cancer Cell Populations & Blood Glucose Level in Prediction of Lung Cancer Patients Survival with Stage III (n=131)
  • 32.
    Weight & Prothrombin Index in Prediction of Lung Cancer Patients Survival with Stage III (n=131)
  • 33.
    Significant Networks betweenClinicopathologic, Biochemic, Hemostasis & Hematologic Data and 5-Year Survival of Lung cancer Patients with Stage III (n=131)
  • 34.
    Results of MonteCarlo Simulation in Prediction of Lung Cancer Patients Survival with Stage III (n=131)
  • 35.
    SEPATH Networks inPrediction of Lung Cancer Patients Survival with Stage III (n=131)
  • 36.
    Holling-Tenner Models ofAlive Supersystem “Lung Cancer-Cytotoxic Cell Population ”
  • 37.
  • 38.
  • 39.
    Conclusions: It wasrevealed that 5-year survival of lung cancer patients with stage III after complete pulmonary resections significantly depended on: 1) lung cancer characteristics; 2) adjuvant treatment; 3) level of blood cell circuit; 4) cell ratio factors; 5) hemostasis system; 6) biochemic homeostasis.
  • 40.
    Address: Oleg Kshivets,M.D., Ph.D. Thoracic Surgeon Department of Surgery Siauliai Cancer Center Tilzes:42-16, 5400 Siauliai, Lithuania Tel. (37041)416614 kshivets@yahoo.com http//:myprofile.cos.com/Kshivets