[Updated August 8, 2014]

#### Types of Predictive Calculators

. In [Full text] [PMID: 19918337] Lowrance (papers) and Scardino (papers) discuss predictive models for prostate cancer covering methods that use risk classifications (e.g. d'Amico risk groups), tables (e.g. Partin tables), risk scores (e.g. CAPRA score), nomograms (e.g. Kattan nomograms)) and systems pathology which refers to using a wider range of variables than the traditional clinical variables (e.g. Aureon Labs' Px method). See [Table 1]. Future methods will likely incorporate genetic inputs to improve prediction accuracy. The authors point out that the need to discretize variables into a number of groups potentially reduces the accuracy of risk classifications, tables and risk scores whereas nomograms do not suffer from this problem (however, whether this potential loss of prediction accuracy is material is not discussed).#### Interpretation

Caution needs to be exercised in interpretation. For example, this article shows what can go wrong by incorrectly assuming that median life expectancy is the amount of time a particular person has left to live.#### Comprehensive Calculators

There are two particularly comprehensive prostate cancer calculator sites plus a range of nomograms on a third site. The links to the first two are easy to remember since their links are nearly the same: http://www.nomogram.org (University of Montreal) and http://www.nomograms.org (Sloan Kettering). The Sloan Kettering site ends in "s" whereas the University of Montreal site does not. The Prostate Cancer Research Institute (PCRI) site contains 9 nomograms, i.e. charts that can be used like calculators.A.

*Memorial Sloan Kettering Calculators*. To access the Sloan Kettering online calculators go to http://www.nomograms.org and then click on Prostate in left hand column. That takes you to a new page and on that click on Open calculator in box in upper right. When new window appears click on "No". The following calculators are provided:

*Pretreatment calculator:*Given PSA, Gleason Subscores, Stage & treatment get prob of disease involvement and disease free progression after 5 years for each of surgery, external radiation and seeds.*Post treatment calculator*: Given PSA, surgical margin and disease involvement get prob of disease free progression in 2, 5 and 7 years.*Hormone refractory calculator:*Given Age, Karnofsy PS, Heomglobin, PSA, LDH, Alkaline Phosphates and Albumin get 1 year, 2 year and median survival probabilities.*Prostate Volume Calculator:*Given dimensions & PSA calculate volume & PSA density.*Life Expectancy:*Given age and race calculate male life expectancy.*PSA Doubling Time*: Given series of dates and PSA values calculate doubling time, slope of log(PSA) vs. time curve and PSA velocity. This calculator does have the restriction that it cannot accept PSA values less than 0.1 which may be a problem if you are using an ultrasensitive PSA assay. In that case enter all your PSA values as 10x or 100x the real PSA value and the doubling time computed will still be correct.

*University of Montreal Cancer Prognostics and Health Outcomes Unit*has a suite of calculators at http://www.nomogram.org. This site is frequently updated with prostate cancer calculators (as well as for other urological conditions) for a wide variety of situations:

*Before Diagnosis*. There are 6 pre-biopsy calculators which can be used prior to biopsy to give the probability of biopsy results. The first is used if PSA is in the 0 - 2.5 ng/ml range, the second is for initial extended biopsy, the third is for extended repeat biopsy, the fourth is for saturation biopsy, the fifth is for 120 day mortability after biopsy based on a comorbidity score and the sixth is for initial sextant biopsy.*Before Treatment I*. There are calculators to predict the probability of Gleason sum upgrade, extra capsular extension, seminal vesicle invasion and lymph node invasion. (Regarding Gleason Score upgrading, note this 2008 paper on factors making Gleason Score upgrade more likely here: [PMID: 18207180] which finds that "A total of 134 patients (50%) were upgraded postoperatively to Gleason score 7 or higher. Preoperative prostate specific antigen greater than 5.0 ng/ml (p = 0.036), prostate weight 60 gm or less (p = 0.004) and more cancer volume at biopsy, defined by cancer involving greater than 5% of the biopsy tissue (p = 0.002), greater than 1 biopsy core (p is less than 0.001) or greater than 10% of any core (p = 0.014), were associated with pathological upgrading. Upgraded patients were more likely to have extraprostatic extension and positive surgical margins at radical prostatectomy (p is less than 0.001 and 0.001, respectively". A second 2008 paper [PMID: 18782303] concluded that "Men with a higher PSA level, perineural invasion and high-volume cancer at biopsy are most likely to be upgraded, while men with a large prostate volume and low-volume cancer at biopsy are more likely to be downgraded. These findings have implications for men with prostate cancer managed without confirmation by RP of their true GS.". Also [PMID: 18778348] concludes that " risk of upgrading is a function of two opposing contributions: (i) a more aggressive phenotype in smaller prostates and thus increased risk of upgrading; and (ii) more thorough sampling in smaller prostates and thus decreased risk of upgrading. When sampled more thoroughly, the phenotype association dominates and smaller prostates are linked with an increased risk of upgrading. In less thoroughly sampled prostates, these opposing factors nullify, resulting in no association between prostate size and risk of upgrading. These findings help to explain previously published disparate results of the importance of prostate size as a predictor of Gleason upgrading.") A 2011 paper by Corcoran et al [PMID: 21895937] analyzed 684 patients with Gleason 6 or 7 on biopsy and found that 50% were upgraded. The upgraded patients had tumors which had significantly smaller tumor volume suggesting that tumor volume can help in predicting an upgrade.*Before Treatment II*. There are calculators for calculating the probability of clinically insignificant prostate cancer, the probability of predominantly transition zone prostate cancer, probability of survival within 30 days of surgery and the probability of 10 year survival.*After Surgery*. There are post op calculators for the probability of PSA recurrence, Local recurrence, Distant recurrence and prostate cancer specific survival.*After PSA Relapse*. There are calculators for the probability of Metastatic progression, mortality for surgical patients undergoing subsequent hormone therapy and mortality after PSA relapse.*Hormone Refractory Prostate Cancer.*There is a calculator for the probability of survival for patients with androgen independent prostate cancer.*Other*. At the bottom of its calculators page the University of Montreal has doubling time and Life expectancy calculators.*Blackberry Calculators*. There are downloadable calculators for the Blackberry. See the May 29, 2010 news item on the New Features page.

C.

*Prostate Cancer Research Institute (PCRI)*. The PCRI has the following nomograms on its site:

- Probability of Extracapsular Extension
- Probability of Seminal Vesicle Involvement
- Probability of Lymph Node Involvement with Tumor
- Probability of Latent or Indolent Tumors of Low Biological Aggressiveness
- Probability of Metastases Five Years After 3D Conformal EBRT
- Probability of Being Disease-Free Five Years After Brachytherapy
- Probability of Median Survival in Castrate Refractory Patients
- Probability of an Abnormal Bone Scan

D.

*SWOP*. A site with several calculators is the http://prostate-riskindicator.com/via.html site, also referred to as SWOP, of The Prostate Cancer Research Foundation is closely related to the Department of Urology of the Erasmus MC, University and Medical Centre of Rotterdam. It has a number of risk indicators. Quoting from the site: "Risk Calculator 1 – the general health calculator is a starting point, looking at family history, age and any medical problems with urination. Risk Calculator 2 – the PSA risk calculator looks at the levels of prostate specific antigen (PSA) in patient’s blood to help predict whether further investigation is required. Risk calculator 3 predicts the chance of a positive sextant biopsy in a man who has never been screened; and also assesses the degree of aggressiveness. Risk calculator 3 + DRE assessment predicts more accurately the chance of a positive sextant biopsy, compared to only assessing a patient’s PSA value (RC 2), but without the necessity of a TRUS. An additional feature is the prediction of a high grade or advanced prostate cancer. Risk calculator 4 is used for men who have previously had PSA screening, but have either had no biopsy or one that was negative. It predicts the chance of a positive sextant biopsy and its degree of aggressiveness. Risk calculator 4 + DRE provides additional information, without the necessity of a TRUS, for assessing men who have previously been screened, whether they have had a prior biopsy or not. It also predicts the chance of a positive outcome and whether that would be high grade or advanced. Risk calculator 5 calculates the chance of having indolent prostate cancer which may not require immediate treatment. Risk Calculator 6 is the latest in the series — it calculates a man’s future risk over the next four years, taking into account age, prostate-specific antigen, digital rectal examination, family history, prostate volume, and previous biopsy status."

#### Other Calculators

.*d'Amico Risk Categories*. Although not a calculator, a useful classification is the d'Amico risk category stratifying disease into Low, Medium and High Risk. More is available in the third paragraph here.

Also check out these calculators:

*Wolfram Alpha*provides a box in which you enter a query and find out where among the population you stand on various medical tests, e.g. enter one of these:

psa 5 age 60

vitamin d 25 age 60 male

bmi 25 age 60 male

life expectancy age 60 male

blood pressure 125/75 age 60 male

ldl cholesterol 125 age 60 male

hdl cholesterol 50 age 60 male

or if you omit the test value then it gives the population reference range, e.g. enter:

psa age 60

*PSA Velocity*. This site provides a raw PSA velocity as well as one adjusted for hemodilution (i.e. taking into account a lower than otherwise PSA value due to dilution in a large amount of blood normally associated with obesity). The PSA Velocity Calculator is also mentioned in this Medical News Today article.

*Risk prior to medical tests*. This calculator from Harvard requires no medical tests as inputs -- only age, height, consumption of animal fat, consumption of tomatos, vasectomy, family history and race. http://www.yourdiseaserisk.harvard.edu/hccpquiz.pl?lang=english&func=home&quiz=prostate

*Risk prior to biopsy*. UT Cancer Therapy & Research Center has a prostate cancer risk calculator where the user inputs race, age, PSA, family history of prostate cancer (yes/no/unknown), digital rectal exam (normal/abnormal/unknown), past biopsy (never/positive/negative/unknown) and get back the probability that a biopsy will show high grade cancer, the probability that a biopsy will show low grade cancer and the probability that a biopsy will be negative for cancer and also show it in a graphical form. (The three probabilities will sum to 100% as they are mutually exclusive.) The calculator is discussed in this medscape article.

*Risk prior to biopsy*. Eric Klein (papers) of the Cleveland Clinic recommends biopsy to his patients if this risk calculator assesses the risk of prostate cancer to exceed 10% in this paper: [Full Text] [PMID: 19652036]. The calculator gives risk of prostate cancer and risk of advanced prostate cancer given: Race, Age, PSA level in ng/ml, Family History of Prostate Cancer, Digital Rectal Examination results, Prior Prostate Biopsy and whether the patient is taking finasteride.

*Risk prior to biopsy*. Given PSA score, DRE, prior biopsy results, race, age and family history, this gives the chance of biopsy finding prostate cancer as well as the chance of Gleason 7 or higher prostate cancer: http://www.compass.fhcrc.org/edrnnci/bin/calculator/main.asp based on [PMID: 16622122] [Full Text]

*Risk prior to biopsy.*Researchers at Memorial Sloan Kettering have [published] and [further validated] a model for assessing the need for a biopsy that they claim is nearly 80% accurate in its prediction of prostate cancer and nearly 90% accurate in its prediction of prostate cancer with Gleason 7 or higher.

*Partin Tables*Prob of disease involvement given PSA, GS and stage. http://urology.jhu.edu/prostate/partintables.php Also see [PMID: 11744442] [Full Text] and [PMID: 17572194] [Full Text]

*Probability of Lymph Node Involvement*. The Yale Formula for the probability that prostate cancer has spread to the lymph nodes is [GS - 5] x [PSA/3 + 1.5 x T], where GS is gleason score, PSA is Prostate Specific Antigen level and T = 0, 1, and 2 for cT1c, cT2a, and cT2b/cT2c. For example, a GS of 7 with a PSA of 6 and staging of cT2c (i.e. 2) would give a (7-5) x (6/3 + 1.5 x 2) = 10% chance of lymph node involvement. In [PMID: 20594769] find that if such involvement is predicted when the formula gives a probability of over 15% then its sensitivity is 39% (i.e. among those whose cancer has spread 39% will have a Yale Formula score of over 15%) and its specificity is 94.9% (i.e. among those whose cancer has not spread 94.9% will have a Yale formula score of less than 15%).

*PCRI calculators.*http://www.prostate-cancer.org/tools/software/software.html

*Radiation Disease Free Probability Excel Spreadsheet.*http://www.prostate-cancer-radiotherapy.org.uk/calculator.htm

*Radiation Oncology Calculators for Palm.*http://radonc.usc.edu/USCRadOnc/Downloadable/PalmOS/PalmPrograms.html

*Life Expectancy without Treatment.*This calculator provides an estimate of life expectancy based on conservative management. It is based on [PMID: 20141675]. Also see this discussion in prostatecancerinfolink.

*Life Expectancy Tables.*These are not online calculators but rather are instructions, a table and a figure.

There is a link to a life expectancy table from the US Social Security Administration (SSA) and an

*explanation*of how to use it on page PROS-A (page 13 of the PDF document) of the NCCN Prostate Cancer Practice Guidelines. They recommend adjusting the ages in the actuarial table to reflect current health status. http://www.nccn.org/professionals/physician_gls/PDF/prostate.pdf

A direct link to the aforementioned

*SSA actuarial table*is here. This table gives total expected lifetime for men of a given age. To get remaning lifetime subtract current lifetime from total lifetime. http://www.ssa.gov/OACT/STATS/table4c6.html

Page SAO-A (page 7 of the PDF document) of the NCCN Senior Adult Oncology Practice Guidelines contains a

*figure with remaining expected years of lifetime*for each 5 year age group as well as upper and lower quartiles. This may be a bit easier to use since it directly gives the remaining lifetime and the quartiles can be used for patients in above average or below average health. http://www.nccn.org/professionals/physician_gls/PDF/senior.pdf The figure is based on [PMID: 11386931] .

*Gait Speed*. This [JAMA 2011] paper (as well as others referenced) have established that gait speed (how fast a person walks is a good predictor of survival. Figure 2 in that paper can be used to predict survival based on age and gait speed.

*Grip Strength*. A 44 year prospective longevity study of 2,239 men [PMID: 21541735] [full free text] found an odds ratio of living to 100 (vs. living to 79 or ealier) of 2.52x for those in highest third of grip strength vs. lowest, 5.75x for those who had never smoked vs. those who had ever smoked, 1.13x for those with highest physical activity vs. least, 2.84x for those with no diabetes vs. those with diabetes, 4.87x for those with normal blood pressure vs. those with elevated blood pressure, 2.26x for those whose mother lived over 60 years (but less than 80) vs. less than 60. There was no relation to paternal longevity.

*Life Span*. The Project Big Life site has a life expectancy calculator based on diet, smoking status and exercise. There are also calculators to estimate how many days in the hospital you will spend over you life time and a salt intake calculator.

*Mortality from Common Diseases*. Four charts giving the mortality from vascular disease, cancer, infection, lung disease, accidents and all causes combined given age, sex and smoking status. There are 2 simple charts: one for men and one for women and two more detailed charts again one for men and one for women. The charts are available [here] as supplements to this paper: [PMID: 18544745]. There is further discussion in [PMID: 18544738] [full text].

*Charlson Comorbidity Score*. Given age and which of 19 diseases the patient has (weighted by association with mortality) the Charlson Score can be calculated and a formula or tables used to calculate the 10 year survival probability given the Charlson score.

You can calculate your Charleson co-morbidity score (higher numbers are

assigned to patients with overall poorer health) using this: Excel spreadsheet. Also see accompanying [Full Text[ [PMID: 15610554] paper.

Now that you have your Charlson score you can input that score into this nomogram which will give separate ten year life expectancies after surgery and after radiation treatment. Also see accompanying paper: [Full Text] [PMID: 17704404].

There are calculators and more info here: [Institute of Algorithmic Medicine Calculator] and here: [info] [Hall et al Calculator] and [Walz 2007]. Also see [PMID: 16770340] . This paper [PMID: 17979925] concluded that clinicians have an accuracy of less than 70% whereas Walz claims that his group's nomogram has an 84% accuracy on a validation sample.

*Probability of Indolent Cancer based on diagnosis variables*. Steyerberg et al published a scoring system in Jan 2007 [PMID: 17162015] (based on updating the original work by Katan et al [PMID: 14532778]) which gives the probability of indolent cancer. It uses the ultrasound and biopsy results (PSA, prostate volume, Gleason score, mm of cancer tissue in cores, mm non-cancer tissue in cores) each of which gives a number which are summed and looked up on a chart showing the probability of indolent cancer. If this probability is high then delay to treatment or active surveillance/watchful waiting could be considered. See page 2 of http://www.pcngcincinnati.org/2007/2007_02.pdf (

*Note:*PCNG Cincinnati site seems to have disappeared.). As noted by Jon Nowlin the original nomogram assumed 6 biopsy cores and could give misleading results if a different number of cores were used. Nowlin has provided an Excel spreadsheet to perform this calculation which corrects for the number of cores [here] and provides explanation of it use [here]; however, a significant caveat is that "Chun et al. [6, 7, 33] demonstrated recently that nomograms developed in the sextant biopsy era may not be able to predict the probability of PCa on needle biopsy in the extended biopsy era, equally accurate as they used to in the sextant biopsy era. In consequence, many clinicians are reluctant to use tools that were developed in the sextant biopsy era [46]." [PMID: 17333203]

*Probability of Cancer in Suspected Patients After Ultrasound*. Focusing on patients with PSA < 10 who are suspected of having cancer and so have had an ultrasound, Garzotto et al (2005) [Full Text] [PMID: 15781880&dopt] develop a [decision tree] model. Unlike scoring systems and nomograms the decision tree is particularly easy to describe. The probability of cancer was less than 5 percent except for the 4 high risk groups illustrated in the decision tree diagram or in words in the abstract. No calculator is really required since the decision tree format is so simple to describe. The variables used are PSA, PSA density (PSAD), existence of hypoechoic lesions, age and prostate volume in cubic centimeters. Existence of hypoechoic lesions and prostate volume are items that are available from the ultrasound. PSA density is PSA divided by prostate volume and therefore also depends on the ultrasound.

*Probability of Recurrence after RP based on diagnosis variables*Cooperberg et al published the CAPRA score system in Nov 2006 [PMID: 17039503] which gives the probability of recurrence after radical prostatectomy based only on variables known at time of diagnosis. It is based on assigning a score to each of 5 risk factors (PSA, Gleason score, clincal T-stage, no. of positive biopsy cores and age) and summing giving a number between 0 and 10 which is looked up on a chart to give the probability of recurrence. See page 3 of PCNG document or this UCSF material. A November 2007 study, [PMID: 17868719], validated the CAPRA score on an independent set of patients from Germany. An older 2003 recurrence table was published by Han et al [online calculator] [Full Text] [PMID: 12544300]. That table was based on a single surgeon at a single institution. Also see the discussion of d'Amico risk categories earlier on this page. We have a separate page where we discuss Biochemical Recurrence.

*Recurrence after Salvage Radiotherapy*Stephenson presented a nomogram for predicting the 6-year progression-free probability after salvage radiotherapy based on prostatectomy PSA, Gleason Score, SV invasion, Extracapsular Extension, Surgical Margins, lymph node mets, persistantly elevated post-prostatectomy PSA, pre-RT PSA, PSADT and radiation dose. See [Figure 3] of [Full Text] [PMID: 17513807]. The nomogram can also be found in slide 15 of his [ASCO 2006 presentation]. In slide 17 he shows that this model predicts better than competing models (actually he shows it has a higher concordance score which is not completely identical but it is suggestive). Also see this video of his presentation [link], his 2004 paper [PMID: 15026399] [Full Text] and his May 2007 paper [PMID: 17513807].

*Probability of Survival in AIPC*. Svatek et al published a nomogram In Jan 2006 to predict the probability of survival in androgen independent prostate cancer. [PMID: 16423446]. It uses the PSA at ADT initiation, PSA doubling time, Nadir PSA on ADT and time from ADT to AIPC diagnosis and gives the 12, 24, 36, 48 and 60 month disease-specific survival probabilities. See page 3 of http://www.pcngcincinnati.org/2006/2006_08.pdf

*Radiation Dosage Calculator*The radiation dosage calculator will estimate your lifetime radiation exposure in mSv given the types and numbers of exposures. This separate calculator will assess the radiation dosage received when flying between specified cities. There is a separate post on radiation risks which has more information on this area.

*Pathology*. A rule of thumb is that for each cubic centimeter (cc) of benign prostate tissue that 0.067 ng/ml of PSA will be produced. Thus for a prostate of 40cc (this is the volume of the prostate, not the volume of the tumor) one would expect a PSA of 40 x 0.067 = 2.68 ng/ml so if the actual PSA were 4.0 ng/ml then there is 4.0 - 2.68 = 1.32 ng/ml that is unexplained and might be due to cancer cells or other factor listed here. [PCRI Insights, August 2008, vol 11.3]. In a December 2008 paper Kato et al devised the following formulas for tumor volume (cc) and percent tumor volume as a function of PSA (ng/ml):

Tumor Volume (cc) = 3.476 + 0.302 x PSA

Tumor Volume (%) = 11.331 + 0.704 x PSA

*Castrate Resistant Prostate Cancer*. This calculator answers the question of whether the patient has castrate resistant prostate cancer and what the optimal treatment is. In association with this calculator readers may wish to view this presentation by Nicholas Vogelzang (papers).

*Body Mass Index Calculator*.The WHO chart shown here (also found in Box 12.3 on page 375 of the WCRF/AICR diet and cancer report and on Wikipedia - also see info on subdivisions) uses height and weight and gives an assessment of underweight/normal/overweight. Chapter 8 of the aforementioned WCRF/AICR diet and cancer report discusses fatness in general and associated risks for various cancers. An alternative to the charts is this BMI calculator.

*Marine Corps Fitness Test.*This [site] describes how to carry out and calculate your fitness using the US Marines Corps Fitness Test.

*Measuring Random Fluctuations Rather than Real Effects.*One caveat regarding the examination of lists of hospitals is that such lists are susceptible to reporting random fluctuations as if they were meaningful confusing such fluctuations with real effects. For example, this link provides a calculator that initially assumes that 100 surgeons in each of 100 hospitals each have a 5% mortality rate among their patients and that a hospital is deemed unacceptable if its surgery rate is 60% higher than the average. Each time you click on Recalculate below the graphic there it does a new simulation showing how many hospitals will be rated unacceptable even though all hospitals are exactly the same. Paradoxically if the death rate assumption is increased to 12% then the number of hospitals deemed unacceptable decreases (!) because there is lesser variation around larger numbers. For more info on hospitals see Choosing a Surgeon Part 2 - Finding a Surgeon

*Non-Medical Calculators*. Although not related to the main topic of our site the following page calculates numerous items in the areas of personal/family (will your marriage last to its Xth anniversary? and more), fun/sports (next move in rock/paper/scissors and more), politics (predict presidential election and more), media (predict the success of a book title and more), health (predict probability of diseases and more) and money (predict how much you will have to retire and more) and economic (predict stock prices and more): [yale calculator links]

## 1 comment:

Enormously helpful--thank you!

Post a Comment