Log in to post a reply
Apr 22, 2018 10:04AM
Apr 22, 2018 04:30PM
Digging a little deeper into this IBIS model I found this at http://www.ems-trials.org/riskevaluator/
Description of breast cancer risk program
The program assumes that there is a gene predisposing to breast cancer in addition to the BRCA1/2 genes. The woman's family history is used to calculate the likelihood of her carrying an adverse gene, which in turn affects her likelihood of developing breast cancer. The risks of developing breast cancer for the general population were taken from data on the first breast cancer diagnosis (ICD-10 code C50) in Thames Cancer Registry area (UK) between 2005-2009. The risk from family history (caused by the adverse genes) is modelled to fit the results in "Familial Breast and Ovarian Cancer: A Swedish Population-based Register Study, Anderson H et al., American Journal of Epidemiology 2000, 152: 1154-1163".
The risk from other classical factors including age at first child and benign disease are combined with familial risk. (my italics)
The latest version of the model (v8) incorporates mammographic density.
Prof. Jack Cuzick
Centre for Cancer Prevention,
Wolfson Institute of Preventive Medicine,
This states familial risk is included, but it never made a difference for those with LCIS in the trials I did, even when 5 close female family members had breast cancer vs. no family history entered. When 7 close female family members with no cancer were entered the risk was no different than if no family history was entered. The 29-year longitudinal study stated family risk was not a factor.
This states benign disease is included, but it never made a difference in bc risk for those with LCIS in the trials I did, whether atypical hyperplasia was checked or not. The 29-year longitudinal study stated family risk was not a factor.
This states age at first child is included, but it never made a difference in bc risk for those with LCIS in the trials I did. I used different ages for first child, but all before age 30, some were childless.
I also included varying ages of first menstrual period from 11 to 17 for those with LCIS--no difference in bc risk; varying ages of menopause from 47 to 54 for those with LCIS--no difference in bc risk. This model does not specifically tout duration of menstruation as a classical factor that is incorporated. The 29-year longitudinal study stated age of menopause is not a factor.
A diagnosis of ovarian cancer at any age made no difference in bc risk for those with LCIS. This model does mention ovarian cancer from a Swedish study.
Using Hormone Replacement Therapy for more than* 6 years in the past with a projection of more than 6 years into the future made no difference in bc risk for those with LCIS. That would indicate you can pop those Premarin pills for a lifetime, ladies, with no impact on your bc risk. Really?
The conclusion that I, a layperson, draw is that LCIS and breast density trump all in the Tyrer-Cuzick IBIS model and other factors have no impact on breast cancer risk. The other inputs appear to be disabled in this program when LCIS is included.
These findings do not support the description of the model that states it incorporates specific classical factors. How can that be? Especially when women with LCIS have no other model to use of course we will think that these classical factors are being included for us when it states they are. How many people take their Saturday night doing dozens of input trials to determine which factors are used and which are not? That's asking too much of the user.
Either the IBIS description needs to be modified to indicate other classical factors are excluded or are immaterial for those with LCIS OR the model is not functioning correctly.
Either this is a valid model which helps inform important life decisions (and it is the only show in town because no other current model incorporates LCIS) or it is too inaccurate to rely upon. We are not using this model as a silly parlor game, nor are the professionals we consult from breast surgeons to genetic counselors, who have used this model with me. Are we supposed to just shrug and figure, "You get what you pay for"? Should we lump this in with other freebie surveys on the Internet? Our medical providers who use this obviously place trust in the inputs and outputs.
* Not to be the nit-picky grammar police, but in the HRT section of the program, it has an option of "more then 6 years" for two separate inputs. It should be "more than 6 years." While mixing up an e and an a does not invalidate a logarithm, it does show a lack of attention to detail and makes me wonder if such lack of attention to detail has permeated the whole damn thing.
We deserve accurate information and I am certain those working on this model wish it to be a useful tool and not a frivolous Internet survey game. Professor Cuzick, or whoever is on the other end of email@example.com is getting an email from me.