Review of last week What is GINA? what does it forbid? what philosophical positions are in conflict? ApoE variants what are they associated with? how big.

Review of last week

What is GINA?what does it forbid?what philosophical positions are in conflict?

ApoE variantswhat are they associated with?how big is the risk of 1 or 2 e4 alleles

compared to general population?(what would be odds ratio?)

is this big enough to change behavior?should you have the right to keep this info pvt.?

what kind of insur. co. is this relevant to?can they ask if you have been tested?

what term refers to this situation?what are potential mkt. consequences?what are pot. consequences if law is changed to

forbid insur. co. from using this info?what ethical/philosophical principles are in conflict?where do you stand?

Patents

What is the social justification for them?Can there be too much or too little patent protection?What are some ways in which Congress/legal system

has tried to strike a balance?What is BRCA1?What was patented?Have there been social costs? Social benefits?What did ACLU argue on behalf of “plaintiffs”?

Do you agree with the following ACLU arguments?If true, should they be cause to invalidate a patent?

it’s hard to ‘invent around’ a gene patentthe patent reduced investigation of mutations

prevalent in some racial groupssome women couldn’t get a ‘second opinion’the tests were more expensive than if there

were no patentMyriad would not allow other companies to

perform tests it was not doing itselfpatenting ‘comparing’ gene sequences

infringes freedom of speech and thought

What did the defendents argue?plaintiff’s don’t have standingNY court does not have jurisdictionplaintiffs shouldn’t be entitled to

‘injunctive relief’ before they infringeUSPTO was just doing what Congress asked it toChakrabarty case and Hand decision

justify patenting living thingsand ‘purified’ natural things

What did the judge decide? Do you agree? Why?

What will happen next? What may be consequencesif decision is upheld or overruled?

Cancer screening – why a controversy over PSA testing?

Natural hx of prostate caPSA testWhat happens when a PSA test is positiveHow many lives (life years) are saved…

per what (# people tested, years of testing, cost…)What are potential benefits and harmsHow do you set a threshold for test “positivity” to try

to balance benefits and harmWhat if harms and benefits accrue to different people

Most common cancers in USA (2009)

incidence deathswomen men women men

1 breast prostate lung lung2 lung lung breast prostate3 colon colon colon colon4 pancreas pancreas5 ovary

>10 cervix cervix

Prostate cancer biology

Who gets it? ~5-10% of 55-70 yr old men develop prostate ca.

over 10 year follow-up period ~50% of 70-80 yr old men who die of other causes have

prostate ca on autopsy

Natural hx of prostate cancer – often (but not always) slow time from onset to symptoms ~10 yrs time from symptoms to death ~ 15 yrs

Will screening identify large # of men with early cancers that would never cause symptoms during their lives? = “excess diagnosis” rate , est. ~ 40% for PSA test

What is done if PSA is positive?bx

What is done if bx is positive (assume no mets)? surgical prostatectomy +/- radiation

complications urinary incontinence (? ~20%) sexual impotence (? ~50%)

Should PSA screening be offered routinely, given 40% excess diagnosis and high complication rate?

Should decision be up to patient?

How should level for “+” test be determined?

PSA = glycoprotein produced mainly by the prostate elevated in prostate cancer and other prostate disorders > 4ng/ml in ~20% of men with prostate ca (sens.) < 4ng/ml in ~90% of men without prostate ca (spec.)

How would specificity and sensitivity change for

>, < 8ng/ml<, < 2ng/ml?

Digression on sensitivity, specificity, etc.

Conditionpositive negative

Test positive TP FP pos. pred. val. TP/(TP+FP)

negative FN TN neg. pred. val. TN/(FN+TN)

sensitivity specificityTP/(TP+FN) TN/(FP+FN)

prevalence = # with disease/total #= (TP+FN) / (TP+FP+TN+FN)

http://en.wikipedia.org/wiki/Sensitivity_and_specificity

Example 1:

203 people screened by fecal blood test for bowel cancer

cancer (confirmed by biopsy)pos. neg.

fecal pos. TP=2 FP=18 pos. pred. val. 2/(2+18)=.1

blood neg. FN=1 TN=182 neg. “ “ 182/(182+1)=.99

sensitivity specificity2/(2+1)=.67 182/(18+182)=.91

prevalence = (2+1)/(2+1+19+182) = 3/203

Example 2: 236 women with BRCA mutations screened annually by 3 modalities. Which is most sensitive?

JAMA 292: 1317 (‘04)How predictive is a + MRI compared to a + mammogram?

A little math: Note 4 #’s (TP, FP, TN, FN) determine 5 ratios:spec., sens.pos. pred. value (ppv)neg. pred. value (npv)prevalence (prev.)

Are any ratios are affected by N = TP+FP+TN+FN?No, so ratios determined by 3 indep. variablesCan express any 2 ratios in terms of other 3

Which ratios are (are not) affected by prevalence?ppv and npv are; (spec. and sens. are not)

Sometimes convenient to express ppv, npv in terms of sens., spec. and prevalence

Choosing a cut-off value for a screening test, e.g. PSAfor prostate ca, affects sens. and spec. of test

Sens. and spec. usually anti-correlated

sens

itivi

ty

1- specificity

specificity 0100100

00 100

xx

xx Receiver-operator charac-

teristic (ROC) curve showingsens. and spec. at differentcut-off values

high cut-off

low cut-off

Fx of time assay is wrong

JAMA 294:66 (2005)

X4ng/ml

ROC for PSA test

*At end of study, all men recommended to have biopsy

For many years, PSA test assumed to be good ideafind and treat the cancer before it spreads

European study ~70,000 men > 50 yo screened by PSA test every 4yrs + 90,000 controls, followed for ~10 yrs

screened controlsprostate cancer rate 8.2% 4.8% “excess” cancers found (8.2-4.8/8.2) = 41%prostate ca deaths 261 363 per 10,000 person yrs 3.5 4.1 relative death rate 3.5/4.1 = .85

confidence interval .65-.98 (p=.04 no diff.)

For every life saved, 1410 men screened48 cancers treated ? ~20 unnecessarily

My rough est. of cost per life year saved from Euro. study

PSA tests 1480 x 2 x $50 148KBiopsies .3 x 1480 x $1000 444KTreatment 48 x $10,000 480K if 10 yrs life gained/life saved ~100K/yr life

# unnecessary treatments (40%) 19# unnecss. cases of urinary incont. 4# unnecc. cases of sex. impotence 10

Do harms outweigh benefits?

How can you decide?

Is it fair to put cost on “harms” and combinethem with other costs -> new cost/QALY?

Should choice of PSA test be left up to patient?

How would you propose to give patientadequate information?

What about patient’s who don’t want to(or can’t) evaluate data?

Add uncertainties of clinical studies …

US study ~32,000 men screened annually, roughlyequal # controls, about 10 yr follow-up period, found no difference in survival (prostate ca mortality slightlyhigher in screened group, 2.0 vs 1.7/10,000 person yrs)

50% of US controls got PSA tests anyway, which may explain smaller “excess” cancer rate in US study (20% vs 40%)

Conclusion – don’t assume screening is always good

Similar issues – weighing benefits vs harms – in screening for other cancers

Next week – breast cancer screeningbalancing benefits vs harmsethical issueswho decidesconsequences of “expert” recommendations

Take-home mid-term on Blackboard, due in 2 weeks

Think about final project; email me as soon as youhave some ideas – I may have helpful suggestions