740 likes | 926 Views
How Will My Patient Do: Nomograms and Predictive Models in Oncology. Peter Scardino, M.D. Chairman, Department of Urology Head, Prostate Cancer Program Memorial Sloan-Kettering Cancer Center New York. What is a Nomogram?.
E N D
How Will My Patient Do: Nomograms and Predictive Models in Oncology Peter Scardino, M.D. Chairman, Department of Urology Head, Prostate Cancer Program Memorial Sloan-Kettering Cancer Center New York
What is a Nomogram? • Statistical definition: scales which you connect with a line, like a slide rule. • Practical definition: A device or model which uses an algorithm or mathematical formula to predict the probability of an outcome, optimized for predictive accuracy.
Established Factors that Characterize Prostate Cancer Clinical stage (TNM) ~ Extent Gleason patterns (1-5) ~ Grade or biologic aggressiveness Serum PSA level (ng/ml) ~ Volume
Clinical Prognostic Factors Each of these clinical prognostic factors (T-stage, grade, PSA) independently predicts pathologic stage and prognosis after treatment. When they are combined, the accuracy of prediction increases.
NOMOGRAM FOR PREDICTING PATHOLOGIC STAGE Johns Hopkins, Baylor, U. Michigan SPOREs PSA 4.1-10 PSA 10.1-20 Gleason Sum Clinical Stage Clinical Stage T1c T2b T3a T1c T2b T3a Organ-Confined 5 69 38 23 57 28 16 (63-74) (32-45) (13-38) (50-64) (22-34) (8-28) 7 48 21 11 36 14 7.2 (42-53) (17-25) (5.8-20) (30-42) (11-18) (4-14) Modified rom Partin AW, Kattan MW, Subong ENP, Walsh PC, Wojno KJ, Oesterling JE, Scardino PT, Pearson JD. JAMA 1997; 277:1445.
NOMOGRAM FOR PREDICTING PATHOLOGIC STAGE Johns Hopkins, Baylor, U. Michigan SPOREs PSA 4.1-10 PSA 10.1-20 Gleason Sum Clinical Stage Clinical Stage T1c T2b T3a T1c T2b T3a Organ-Confined 5 69 38 23 57 28 16 (63-74) (32-45) (13-38) (50-64) (22-34) (8-28) 7 48 21 11 36 14 7.2 (42-53) (17-25) (5.8-20) (30-42) (11-18) (4-14) Modified rom Partin AW, Kattan MW, Subong ENP, Walsh PC, Wojno KJ, Oesterling JE, Scardino PT, Pearson JD. JAMA 1997; 277:1445.
PSA Progression-free Probability after RP for cT1-3a Prostate Cancer by Pathological Stage Confined ECE SVI LN+
0 10 20 30 40 50 60 70 80 90 100 Preoperative Nomogram for Prostate Cancer Recurrence Points (31) (47) (57) = 135 pts PSA 4 20 0.1 1 2 3 6 7 8 9 10 12 16 30 45 70 110 T2a T2c T3a ClinicalStage T1c T1ab T2b Organ confined = 21% 2+3 4+ 3+ 2 Biopsy Gleason Grade 2+ 2 3+3 3+ 4 135 pts Total Points 0 20 40 60 80 100 120 140 160 180 200 47% 60MonthRec. Free Prob. .96 .93 .9 .85 .8 .7 .6 .5 .4 .3 .2 .1 .05 Instructions for Physician: Locate the patient’s PSA on the PSA axis. Draw a line straight upwards to the Points axis to determine how many points towards recurrence the patient receives for his PSA. Repeat this process for the Clinical Stage and Biopsy Gleason Sum axes, each time drawing straight upward to the Points axis. Sum the points achieved for each predictor and locate this sum on the Total Points axis. Draw a line straight down to find the patient’s probability of remaining recurrence free for 60 months assuming he does not die of another cause first. Note: This nomogram is not applicable to a man who is not otherwise a candidate for radical prostatectomy. You can use this only on a man who has already selected radical prostatectomy as treatment for his prostate cancer. Instruction to Patient: “Mr. X, if we had 100 men exactly like you, we would expect between <predicted percentage from nomogram - 10%> and <predicted percentage + 10%> to remain free of their disease at 5 years following radical prostatectomy, and recurrence after 5 years is very rare.” • 1997 Michael W. Kattan and Peter T. Scardino Kattan MW et al: JNCI 1998; 90:766-771.
Patient population: validation of preoperative recurrence nomogram USA Cleveland Clinic n= 1168 pts. LSU n= 583 pts. UCLA n= 617 pts. USC n= 1501 pts. Europe Hamburg n= 1134 pts. Rotterdam n= 475 pts. Australia Sydney n= 754 pts. Total n= 6232 pts. Concordance index 0.751
Urologists vs. Preoperative Nomogram • 10 case descriptions from 1994 MSKCC patients presented to 17 urologists • In addition to PSA, biopsy Gleason grades, and clinical stage, urologists were provided with patient age, systematic biopsy details, previous biopsy results, and PSA history. • Preoperative nomogram was provided. • Urologists were asked to make their own predictions of 5 year progression-free probabilities with or without use of the preoperative nomogram. • Concordance indices: • Nomogram = 0.67 • Urologists = 0.55, p<0.05
Progression-free probability by risk group Intermediate risk Low risk T2b or Gl 7 or PSA 10-20 * T1c/T2a, Gl 2-6, PSA <10 High risk T2c or Gl 8-10 or PSA >20 * * * p < 0.05 Mod. From D’Amico et al JAMA 280:969-74, 1998
From Carroll and Kattan, based on the CaPSURE Database, 2002.
Palm Pilot nomogram software program: algorithm for prognosis and pathologic stage • Includes pretreatment and postoperative predictions. • Uses published nomograms in prostate cancer. Nomogram Available at http://www.mskcc.org/prostate/nomograms
Advantages of nomograms: interpreting discordant results • A 59 y.o. man has a poorly differentiated (Gleason grade 4 + 5 = 9) cancer in 1 biopsy core taken from an impalpable (cT1c) prostate because of a PSA of 6. How likely would surgery control this cancer? 83% 63% 43% 23% 3% • Answer: 83% (37% confined, 40% ECE, 15% SVI, 8% LN) Nomogram • A 53 y.o. man had a PSA 47 when first examined. There was a firm palpable nodule in the prostate (cT2a). Needle biopsy showed Gleason grade 3+3=6 cancer in 3 of 6 cores. Is this a “curable” lesion? What is the 5 yr. freedom from recurrence?88% 73% 58% 43% 28% • Answer: 58% (22% confined, 60% ECE, 10% SVI, 8% LN)
Nomograms in Oncology • Prostate cancer • Kidney cancer • Sarcoma • Gastric cancer • Breast cancer • Melanoma • Colorectal carcinoma • Pancreatic cancer • Lung cancer
Renal Cancer Prognosis: Postoperative Nomogram_________________________________________________________________ Kattan M., Motzer R., Reuter V., and Russo P. Department of Urology Memorial Sloan-Kettering Cancer Center New York, NY J. Urol. 2001
Postoperative Prognostic Nomogram for RCC Instructions for Physician: Locate the patient’s symptoms (I=incidental, L=local, S=systemic) on the Symptoms axis. Draw a line straight upwards to the Points axis to determine how many points towards recurrence the patient receives for his symptoms. Repeat this process for the otheraxes, each time drawing straight upward to the Points axis. Sum the points achieved for each predictor and locate this sum on the TotalPoints axis. Draw a line straight down to find the patient’s probability of remaining recurrence free for 5 years assuming he or she does not die of another cause first. Instruction to Patient: “Mr. X, if we had 100 men or women exactly like you, we would expect between <predicted percentage from nomogram – 10%> and <predicted percentage + 10%> to remain free of their disease at 5 years following surgery, though recurrence after 5 years is still possible.” Kattan et al, J. Urol, 2001
Case History #2: 68 yo female presents with tumor found incidentally during ultrasound for gallbladder pain. Undergoes partial nephrectomy for 4cm chromophobe RCC.
Case History #3: 62 yo male presents with painless gross hematuria and undergoes resection of 10cm conventional clear cell carcinoma invading IVC.
Predictive Nomogram: Benefits __________________________________________________ • Patient counseling. • Equal risk stratification in clinical trials. • Entry of patients with poor prognosis into adjuvant clinical trials.
Prostate Cancer Disease States:Nomograms Predict Transition Rates Between Disease States Cancer suspected (PSA, DRE) Localized cancer Rising PSA Metastases Post-castrate mets Death, other Death, cancer Salvage Rx Modified from Scher et al Urology 2000
PREDICTIVE MODELS FOR PROSTATE CANCER • Patient counseling • Follow-up surveillance scheduling • Rational application of adjuvant therapy • Risk stratification for clinical trials
Prostate Cancer Disease States • Biopsy negative. • Need repeat biopsy? • Repeat biopsy nomogram Cancer suspected (PSA, DRE) Localized cancer Rising PSA Metastases Post-castrate mets Death, other Death, cancer Salvage Rx Modified from Scher et al Urology 2000
Repeat biopsy: multivariable analysis Lopez-Corona et al., In Press, J. Urol
Repeat biopsy nomogram ≤ 1/2 lobe N >1/2 lobe Y N Y N PSA Slope Y N Prob. +(Bx) 12 Mo. from first neg Bx Prob. +(Bx) 24 Mo. from first neg Bx Prob. +(Bx) 36 Mo. from first neg Bx
Prostate Cancer Disease States Cancer suspected (PSA, DRE) Localized cancer Rising PSA Metastases Post-castrate mets Death, other Death, cancer • Needs aggressive treatment? • Indolent cancer nomogram Salvage Rx Modified from Scher et al Urology 2000
Indolent prostate cancer • total tumor volume <0.5cc • confined to the prostate (no focal or established extracapsular extension) • no Gleason pattern 4 or 5 in RP specimen
Indolent prostate cancer dataset • RRP between August 1, 1986 and March 1, 2000 • Baylor College of Medicine (n=572) • University Hospital Hamburg-Eppendorf (n=450) • All men had to have systematic needle biopsy (> 6 cores) • Exclusions: • pretreatment PSA > 20 • primary or secondary Gleason grade 4 or 5 cancer in biopsy • positive cores > 50% • total cancer in biopsy cores > 20mm • benign tissue in all cores < 40mm • Left 409 men for analysis (indolent N=80, 20%)
Base indolent nomogram Pre.Tx.PSA Pri.Bx.Gl Sec.Bx.Gl Prob. Indolent Ca.
Full indolent nomogram Pre.Tx.PSA Clin. Stage Pri.Bx.Gl Sec.Bx.Gl U/S Vol mm Cancer mm nonCa Prob. Indolent Ca.
Prostate Cancer Disease States Cancer suspected (PSA, DRE) Localized cancer Rising PSA Metastases Post-castrate mets Death, other Death, cancer • What are the pathologic features of the cancer? Salvage Rx Modified from Scher et al Urology 2000
Predicting Pathologic Stage • Organ confined – DRE, PSA, biopsy Gleason grade, %cores +, mm cancer, mm non-cancer • Side of extracapsular extension – side-specific DRE and biopsy results, PSA • Seminal vesicle invasion – DRE, PSA, biopsy Gleason grade, % cancer in cores at the base • Lymph node metastases – DRE, PSA, biopsy Gleason grade, your % LN+.
Prostate Cancer Disease States Cancer suspected (PSA, DRE) Localized cancer Rising PSA Metastases Post-castrate mets Death, other Death, cancer • How well will each treatment work? Salvage Rx Modified from Scher et al Urology 2000
Preoperative Nomogram for Prostate Cancer Recurrence 0 10 20 30 40 50 60 70 80 90 100 Points PSA 4 20 0.1 1 2 3 6 7 8 9 10 12 16 30 45 70 110 T2a T2c T3a ClinicalStage T1c T1ab T2b 2+3 4+ 3+ 2 Biopsy Gleason Grade 2+ 2 3+3 3+ 4 Total Points 0 20 40 60 80 100 120 140 160 180 200 60MonthRec. Free Prob. .96 .93 .9 .85 .8 .7 .6 .5 .4 .3 .2 .1 .05 Instructions for Physician: Locate the patient’s PSA on the PSA axis. Draw a line straight upwards to the Points axis to determine how many points towards recurrence the patient receives for his PSA. Repeat this process for the Clinical Stage and Biopsy Gleason Sum axes, each time drawing straight upward to the Points axis. Sum the points achieved for each predictor and locate this sum on the Total Points axis. Draw a line straight down to find the patient’s probability of remaining recurrence free for 60 months assuming he does not die of another cause first. Note: This nomogram is not applicable to a man who is not otherwise a candidate for radical prostatectomy. You can use this only on a man who has already selected radical prostatectomy as treatment for his prostate cancer. Instruction to Patient: “Mr. X, if we had 100 men exactly like you, we would expect between <predicted percentage from nomogram - 10%> and <predicted percentage + 10%> to remain free of their disease at 5 years following radical prostatectomy, and recurrence after 5 years is very rare.” Nomogram • 1997 Michael W. Kattan and Peter T. Scardino Kattan MW et al: JNCI 1998; 90:766-771.
Addition of Percent of Cores Positive to Nomogram AUC unchanged, at 0.75
Existing Nomograms for Prostate Cancer • Pathologic stage • Radical prostatectomy: preoperative (5 year) • 3D conformal external beam radiation therapy • Biochemical progression (5 year) • Metastases (7 year) • Brachytherapy (5 year) • Radical prostatectomy: postoperative (7 year)
Prostate Cancer Disease States Cancer suspected (PSA, DRE) Localized cancer Rising PSA Metastases Post-castrate mets Death, other Death, cancer • Improved predictions Salvage Rx Modified from Scher et al Urology 2000
Postoperative Nomogram for PSA Progression after Radical Prostatectomy C-Index: 0.80 From Kattan MW, Scardino PT et al. J Clin Oncol 1999
WHY A NEW POSTOPERATIVE NOMOGRAM? • Improved prognosis of prostate cancer in the recent era independent of stage, grade, PSA • Cagiannos et al. J Urol 2004 • Patients receiving adjuvant radiation therapy were considered treatment failures in the original nomogram • Nomogram predicts the 7-year PFP many patients remain at risk after 7 years • 10-year PFP is optimal endpoint for biochemical progression as the risk of progression after 10 years is 3% (95% CI, 0-7%) • Changing prognosis based on disease-free interval status • A nomogram that recalculates patient’s prognosis for a disease-free interval that he has achieved would be useful for patient counseling and follow-up scheduling
CHANGING PROGNOSIS BASED ON DISEASE-FREE INTERVAL MAINTAINED * As defined by PSA < 0.2 ng/mL within preceding 12 months
PATIENTS AND METHODS • Cox proportional hazards regression analysis model the pre- and postoperative clinical data of 2055 pts with clinically localized prostate cancer who underwent radical prostatectomy by 1 of 2 surgeons between 1983 and 2003 without prior therapy • Variables included in nomogram: • Patients with null values excluded 1881 eligible patients * Treated as time-dependent covariate
ENDPOINT • Disease progression: • PSA 0.4 or greater and a confirmatory rise (time of failure is date of 2nd PSA rise) • Initiation of androgen-deprivation therapy after RP • Initiation of external-beam radiation therapy > 1 year after RP • Clinical disease recurrence • Death from prostate cancer • Patients censored if lost to follow-up or death from causes other than prostate cancer
NOMOGRAM VALIDATION • Nomogram applied to independent cohort • 2103 patients who underwent radical prostatectomy at MSKCC between 1985-2003 who were not included in modeling cohort • No patient received prior therapy before RP • Patients censored at the time of adjuvant radiation therapy