Predicting breast cancer

leaf

This 2006 editorial said the study found that the Gail model worked well for predicting how many women in a population group got breast cancer, but it did only slightly better than a coin toss in predicting which *individual* in a population will get breast cancer.



The Gail model was developed for use in the US, and this study examined how well it predicted breast cancer in Florence, Italy, and compared it to an Italian model.





It gives me even less confidence in the Gail model. :-(, or any current model.
But it gives me some more information how breast cancer risk is assessed.



I would have felt better if my health care providers told me how uncertain we are about predicting breast cancer in individual patients.




It seems the Gail model is mainly useful to predict health care needs-how many people in a population (at least in the US and Florence, Italy) will get breast cancer.





(For a random coin toss, the prediction would be 0.5 =50%. Complete accurate prediction = 1=100%. These models gave about 0.59=59%, only slightly better than a coin toss.)








" It’s tough making predictions, especially about the future.
— Attributed to many individuals, including Yogi Berra. But, as Yogi
himself said, “ I really didn’t say everything I said. ” ( 1 )





The best-known model for predicting an individual woman’s
chance of being diagnosed with breast cancer is the Gail model
( 5 , 6 ) . This model includes the following risk factors: current age,
race, age at menarche, age at fi rst live birth, the number of fi rst-
degree relatives with breast cancer, the number of previous breast
biopsy examinations, and presence of atypical hyperplasia. The
model predicts a woman’s likelihood of having a breast cancer
diagnosis within the next 5 years and within her lifetime (up to age
90 years). This and similar risk prediction models are readily avail-
able to clinicians and patients around the world through the Inter-
net ( 7 , 8 ) . A version of the Gail model available on the National
Cancer Institute’s Web site ( http://www.cancer.gov/bcrisktool/ )
is viewed 20 000 to 30 000 times each month (Rehmert JH: per-
sonal communication).



The Gail model was developed and validated in the United
States ( 5 , 6 , 9 – 11 ) . However, breast cancer incidence rates vary
fourfold by geographic location, with some of the highest rates in
the United States and northern Europe ( 12 ) . Given this wide vari-
ation in breast cancer incidence, it was not known how well the
Gail model would perform internationally. In this issue of the
Journal , Decarli et al. ( 13 ) used data from a multicenter case –
control study in Italy and from Italian cancer registries to develop
a new breast cancer risk prediction model that used the same risk
factors as the Gail model. Decarli et al. then tested the relative
predictive accuracy of the Italian and Gail models by using inde-
pendent data from a cohort study in Florence, Italy ( 14 , 15 ) . They
found that the two models produced similar results.




Cancer risk predication models are commonly assessed in two
ways: by measuring their performance at the population level and
at the level of the individual woman. Decarli et al. assessed each
model’s performance at the population level by comparing the
number of women in their study who the model estimated (E)
would develop breast cancer with the number of women who
actually were diagnosed with breast cancer (observed [O]). The
Italian and Gail models estimated that 186 and 180 women, re-
spectively, would develop breast cancer. The actual number was
194. Therefore, the overall E/O ratios for the Italian and Gail
models were similar (0.96, 95% confi dence interval [CI] = 0.84
to 1.11, and 0.93, 95% CI = 0.81 to 1.08, respectively).



Decarli et al. also assessed each model’s performance at the
level of the individual woman. A model that discriminates well at
this level should consistently predict a higher risk of breast can-
cer for women who will be diagnosed with the disease than for
women who will not. Decarli et al. randomly selected pairs of
women, one of whom was diagnosed with breast cancer and one
of whom was not, to determine the frequency with which each
model calculated a higher risk for the woman who developed
breast cancer. The resulting calculation produced a concordance
statistic, whose value could range from 0.50 (equivalent to a coin
toss) to 1.0 (perfect discrimination). The concordance statistics
for the Italian and Gail models were essentially the same, ap-
proximately 0.59 (with 95% confi dence intervals that ranged
from 0.54 to 0.63). In other words, for 59% of the randomly se-
lected pairs of women, the risk estimated for the woman who was
diagnosed with breast cancer was higher than the risk estimated
for the woman who was not. Unfortunately, for 41% of the pairs
of women, the woman with breast cancer received a lower risk
estimate than her cancer-free counterpart. Thus, for any given
woman, the two models were better at prediction than a coin
toss — but not by much.



Figure 1 illustrates the problem of a similarly low concor-
dance statistic (0.58) noted by Rockhill et al. when they applied
the Gail model in the Nurses’ Health Study ( 10 ; Rockhill Levine
B: personal communication). There is no place along the x -axis
where one can clearly separate the group of women with breast
cancer from the group without. If, instead of proportions, the
fi gure used actual numbers of women, the curve for the 80 755
women who did not develop breast cancer would swallow up the
far smaller group of 1354 women who did. Rockhill et al.
also reported a sensitivity of 0.44 and specifi city of 0.66 using
a 5-year risk of developing breast cancer of 1.67% as their cut
point . This result illustrates a conundrum that all clinicians know:
many “ low-risk ” women develop breast cancer, while many
“ high-risk ” women do not. In contrast, Fig. 2 illustrates what
happens when a fi ctitious risk prediction model clearly
differentiates between the two groups and has a concordance
statistic of 1.0.



Recent attempts to improve the Gail model by adding infor-
mation on other risk factors, such as breast density, have im-
proved the concordance statistic somewhat by bringing it up to
0.66 ( 16 , 17 ) . However, in most situations, even a concordance
statistic of 0.66 is still too low to make management decisions for
individual patients.



Why is it so diffi cult to develop worthwhile breast cancer pre-
diction models for individuals? First, the risk factors used in cur-
rent models are widely prevalent throughout the population and
are neither highly sensitive nor highly specifi c. In addition, a risk
factor must be very strongly associated with a disease (with a
relative risk of about 200) to be worthwhile for screening ( 18 ) ,
and the same appears to be the case for accurate prediction using
combinations of risk factors. Most risk factors for breast cancer
are relatively weak. Even “ strong ” risk factors, such as older age,
mammographically dense breasts, and radiation exposure, are as-
sociated with relative risks of less than 10. [Deleterious BRCA1
mutations in young women may be an exception ( 19 ) .]



Current breast cancer risk prediction models perform well for
populations but poorly for individuals. Cancer risk information,
now readily available through the Internet, can show an indi-
vidual that she is a member of a group that is at higher risk for
a cancer diagnosis compared to the average population. This
is valuable information, but it must be interpreted carefully
( 10 , 20 , 21 ) . Both clinicians and patients must understand the nu-
meric information resulting from breast cancer risk prediction
models in order to use them effectively ( 20 , 22 ) . Because we still
cannot predict accurately enough which individual woman will
or will not develop breast cancer, there is much work yet to do in
the fi eld of cancer risk prediction.

http://jnci.oxfordjournals.org/cgi/reprint/98/23/1673

TenderIsOurMight

Thank you for posting this Leaf. It is greatly informative, and a topic which I think of quite a bit when I urge my daughter and her friends to eat healthy, stay fit, and threw a bit of a fit when an indiscriminate doctor did an unnecessary hip xray for running pain without Mom's approval.



I wonder if gene profiling will replace Gail in the long term. Of course, laws would need to be in place about genetic discrimination and insurance first. In this case, science may precede the use...



I actually thought my risk for bc was high. My doctor though it was low.



Thank you for a great post.

Tender

Beesie

I'm actually surprised that the Gail model is even good at predicting how many women will get breast cancer.  It seems that pretty much everyone gets a really high risk % from the Gail model, much higher than what their doctors tell them when looking at their detailed personal history.  The factors used by the Gail model have always struck me as being too simplistic and incomplete.  JMO.

Indigoblue

Predicting the unpredictable is like wonderinf  if it will rain on your wedding day.  Who's Gail, anyway?

just kidding...

Hi Leaf, hope you're doing well.  Considering the "american" diet and the environment, without the genetic components to combat cancer cells, we are vulnerable no matter what measures we take in hopes not to get some form of cancer. 

Look around.  I go for walks late at night; the odor of carbon monoxide, dryer sheets, cleaning fluids, noise and air polution is everywhere, as I walk beneath the zip and the zap of electric wires, the odor of gas leaks, and wonder why mankind continues to chop down every little square of available land.

The creeks, streams, rivers, and local streets are cluttered with trash and detergents, oils, and who knows what?'

Can't buy anything that isn't encased by, with, or wrapped in some form of plastic by-product. 

Predicting cancer is easier than predicitng weather, since we know about what is happening to the planet, and most prefer to ignore what they know is truth.

Indi

Beesie

Indi, it's good that you are taking care of your health by going for walks.  There's nothing like the smell of fresh air..... oops..... maybe not!      Actually, I loved the description of what you encounter on your walks.  It's so true. 

The things you mention are the sort of factors that the Gail model - and any model - can't account for.  Personally I believe that we may be born with a higher or lower propensity to get various cancers (and other diseases) but then it's a series of internal (eg. stress, diet) and external (eg. pollution) environmental factors that trigger whether we actually get a disease or not.  I also believe that the factors that might trigger BC in one woman may have no effect at all on another woman.  That's why one woman can eat the best possible diet but still get BC, while another can gorge herself on junk food, fatty meats and alcohol, and never get BC. 

All that said, it can't be denied that medical science is better today than it used to be at predicting risk.  And new developments continue.  Similarly, BC is being diagnosed earlier and earlier, requiring less invasive and fewer systemic treatments.  And treatments are getting more and more targetted.  We may get some benefit; hopefully the next generation, our daughters and nieces, will face an entirely new and improved world when it comes to BC prediction, diagnosis and treatment.

leaf

I think we simply do not know how to predict bc. Why does the person who eats junk food, doesn't exercise, stressed out, and every single risk factor you can list (except for BRCA) not get breast cancer, and the woman who exercises, handles her weight and stress, etc get bc? I think one point of the paper is that there is NO risk factor for breast cancer that is specific enough to make any treatment choices (except perhaps BRCA.)



If doctors are NOT using the Gail model, how are they coming up with numbers????? Out of the air?



I am very bummed out, because I have LCIS + ALH, a weak family history, and am interested in PBMs for my LCIS.



The NCI website specifically excludes LCIS in their modified Gail model calculations. http://www.cancer.gov/bcrisktool/ . If I put my numbers in the Gail model, pretending I only have ALH, I get 23.1%.



But I would like my PBM decision based in some part on science. People will probably do something different for a risk of 10% than a risk of 40% or 60% or 80%.



I have been given estimates of my breast cancer risk as

a) 10-60% by a major university, with the figure of 10-20% if they had to choose a number. (7-07)

I am almost CERTAIN they did NOT put my data into the Gail model. I think they used the notorious ACS MRI guideline paper, which IMHO is internally inconsistent with respect to LCIS. (The major university nurse practitioner said they use the Gail model to predict breast cancer, and I finished her sentence by saying 'Which the NCI website specifically excludes." (because of my LCIS.))



b) 40% by the same major university board certified genetics counselor (3-06). This was before my

2 negative breast biopsies in 2-07.



c) The 'notorious' Dr. Halls' calculator gives me between 50-83% lifetime risk, depending how you figure my sporatic birth control tablet use and if I take tamoxifen.http://www.halls.md/breast/risk.htm



and

d) maybe 20-30% by my onc (who gets interrupted every 3 minutes.)



This editorial suggests the Gail model is junk for making decisions about an individual.



This makes it quite difficult to decide on a rational basis whether or not to have PBMs. It puts every estimate I have been given of my bc risk into serious question. I would rather they communicate to me that their estimates are really no better than a coin toss for each individual.



Why don't docs say this?

yellowrose

According to the Gail model, I was low risk even though I am 5th in my family with BC.  I had no 1st degree family members with BC so the Gail model didn't apply.  The Klaus model takes in 2nd degree relatives and my bc risk jumped. 

After being diagnosed, I visited a genetic counselor who told me they also look at other types of cancer that lead to a higher risk of BC, like colon cancer.  That added to my family history and I ended up with a 43.8% rating and am now awaiting my BRCA results.  The counselor said even if its negative, it could very well be genetic due to my high risk %. 

I would just like the cancer to stop with me. I would hate to see my son or future grandchildren fighting this battle.  But at least if they know the risk, they can be more proactive with health issues.

KarenC

I also saw this Gail model and took the test-my risk was slightly higher as I have been on birth control pills for 15 years to treat ovarian cysts and I have never had any children-I'm 42.

However, it was interesting about the biopsy question-have you had more than one?

Yes and have they been benign?

Yes

Supposedly one of the higher numbers for me came from this as they say the risk is higher for the reasons I have had to have the biopsies but the first one was a fatty tumor-lipoma when I was 34-I had a lumpectomy to remove it and surrounding tissue.

The second, more recent in August was microcalcifications gain on my right side-stereotactic biopsy and also benign.

So two different reasons why I had biopsies-I'm not sure I'm any higher risk then someone with fibrodenomas and cysts other than my age, no childbirths and being white-as it did ask race/ ethnicity in the risk assessment.

Karen

Indigoblue

Interesting data, especially when we're apt to apply it to genetic/nongenetic, estrogen/estrogen negative, women with no family history available, women with a family history of heart disease or diabetes, etc., in which no family is around to ask what the history may or may not imply.  The genetic screening is not usually done unless there is a clear background of b.c. 

I am a perfect candidate for estrogen positive, her2 positive, progesterone positive cancers...which may evolve, who knows. At present, my cancer is triple negative.  I have every single A plus score for having estrogen positive cancer, however, grade 3, Stage I, triple neg, aggressive invasive ductal carcenoma was the first to present itself.  Who knows, had the radiologist and obgyn found it sooner, DCIS would have expelled it, and maybe some other form of LCIS or metaplastic mutating cells would have eventually shown up.

My mother had Inflammatory Breast Cancer, with a mammogram showing nothing only 3 WEEKS previous to her adventures with diagnosis, prognosis, surgery, treatment.  Cancer is usually going to appear sooner or later in one's life if they live long enough; or so the surgeons and oncologists tell me so.

So much is unknown.  There is so much I want to know.  There must be a connecting factor somewhere in the structure of cells and enzymes, proteins, DNA, and genetic fragility.  Dang! I truly wish I was a genius, an Einstein, a "big brain", and could come up with the truth, the answer, the tiny little molecule causing all this misery.

Saw Bobby Dylan tonight...:  "The answer my friend, is blowing in the wind, the answer is blowing in the wind..."

Indi