Sunday, March 27, 2011

A genetic predisposition: from speculation to opinion to 'fact' without any data (Part II)

In my last blog post I presented the ‘evidence’ (or lack thereof) on which the theory of a genetic/constitutional predisposition for Asherman's syndrome was based. To recap the gist of these observations which pass for proof:

a) some women develop a severe form of IUA after undergoing the ‘same’ traumatic procedures as others who do not acquire AS, and 

b) some women respond more ‘favourably’ to treatment than others who suffer from recurrent adhesions. 

Before deconstructing these observations, the most obvious argument against a genetic basis is the lack of familial clustering of the ‘pathology’. Assuming a polygenic mode of inheritance (as opposed to a Mendelian one) it would be expected that those with a first degree relative with Asherman’s syndrome would have a higher risk of  developing AS than those without a close relative who has the condition. I have yet to read a study of identical twins with Asherman’s syndrome, let alone any study showing the sisters or daughters of patients having a higher risk. In over a century since its first report, there is no evidence of familial clustering to lend support to this theory. Nor is there any evidence that women with scarring defects (eg. keloids) or connective tissue disorders (Marfan’s syndrome, Ehlers-Danlos syndrome) are more prone to AS as one might expect. 

How can trauma be quantified when surgery is blind?

As for the ‘evidence’ above, with regards to a), what proof is there that the women who develop severe IUA actually underwent the exact same trauma as those who did not acquire it? This is a an unfounded statement given differences in doctor’s skill, technique and more importantly, the mis- or under-diagnosis of AS and the blind nature of D&Cs- the number one cause of AS. The problem with bind surgery is that one relies on guess work and instinct to not scrape too deeply. ‘Too deeply’ can be one millimeter too much. How is it possible to detect this difference when not even an ultrasound is used during the procedure? Even with visualization it is not possible to tell where the functional endometrium ends and the basal endometrium begins. However, techniques which utilize visualization for intrauterine surgery (i.e. hysteroscopy) are inherently less risky with regards to instrumental injury because they allow the surgeon to only scrape/dissect/remove parts of the endometrium which need to be treated thus sparing underlying and adjacent tissue from potential injury. Furthermore, each woman has a different anatomy, and different location of retained products of conception. Some have retroverted uteri, others have Mullerian malformations, and the shape, widths and lengths of uteri vary enough between women to be a significant factor with regards to acquiring an injury during a blind procedure. A D&C consists of blindly scraping away the top layer of the endometrium (the functional layer) which can vary in thickness between women but is often not thicker than a few millimeters in most parts. It is simply impossible for a clinician or midwife to know whether they have scraped into the basal endometrium, particularly when no visual guidance is used. 

In summary, the above argument is akin to saying that not all smokers develop lung cancer therefore lung cancer is not caused by cigarettes. Why not theorize that those who develop lung cancer have a genetic predisposition instead? (Note: this comment was intended to highlight the absurdity of believing that cigarette smoking does not cause lung cancer, and hence the absurdity of believing that the fact that all women who have D&Cs do not acquire Asherman's syndrome proves that the condition is not caused by D&Cs but is instead genetic. I later learned that the famous statistician RA Fisher actually theorized half a century ago that people who are genetically predisposed to lung cancer are also genetically predisposed to becoming smokers. Of course this theory was refuted by a study of monozygotic twins versus dizygotic twins who were discordant for cigarette smoking, and it turns out,  lung cancer risk. This example points out how easily theories can be taken seriously just because an expert, or in this case, a genius put it forth, and that experts too have biases which affect their views (he smoked and was a consultant for the tobacco industry!). These theories, will not hold up against evidence from well-designed studies, but until these are done, they impede medical progress).

Recurrence depends on initial severity and treatment methods

With respect to the second part of the argument, the fact that some women respond more ‘favourably’ to treatment than others who suffer from recurrent adhesions can be rationally explained in other ways.

Firstly, it is not surprising that some women need more than one hysteroscopic adhesiolysis surgeries than others, depending on severity. This has been known for many years and reported in the literature. The more damage incurred on the endometrium, the higher the rate of IUA recurrence, and the more surgeries needed for correction, although at some point, damage is too great to allow for endometrial regeneration resulting in recurrent adhesions or fibrotic endometrium.

Another important point to consider is that hysteroscopy was not used for diagnosis or treatment when the observation about treatment outcomes was made. Asherman’s syndrome was still ‘treated’ with blind D&C, the same procedure which is responsible for most cases of AS. This would understandably result in inconsistent outcomes. The absence of accurate diagnosis (and hence classification) in those days also distorted the correlation between severity and outcome. A patient with seemingly severe AS presenting with total amenorrhea may ‘inexplicably’ have had a better outcome after surgery than someone with supposedly less severe AS who had some menstrual flow. We now know that amenorrhea is not always due to severe and widespread adhesions. It can be due to cervical adhesions alone. As the rest of the uterus may be intact in such cases, the reproductive prognosis is better than a patient who has deep adhesions in part of her uterus but no outflow obstruction (and thus has menstrual bleeding, albeit reduced flow). 

With regards to treatment, estrogen therapy was not carried out as part of post surgical therapy in the past. It is generally accepted that estrogen supplementation plays an important role in the prevention of adhesion recurrence following surgery/adhesioloysis by stimulating endometrial regeneration. Thus in the absence of hysteroscopic adhesiolysis, uterine stents and estrogen therapy, more severe cases were less treatable than they are today. 

It is well known that the functional endometrium is regenerated from the underlying basal endometrial layer. As a simple analogy, imagine the lining of the uterus as a lawn. The grass seeds would be the stem cells from which the grass grew with the addition of water (i.e. estrogen). The functional layer would be the grass, and the roots would be the basal endometrium. If someone came along to trim the lawn (i.e. D&C) and accidentally dug out the roots of the grass, the grass would regrow after watering only in the areas where there were either some residual roots, or seeds. Imagining that the seeds are in a slightly deeper layer of soil (for the purpose of this analogy), if the gardener dug very deeply so that even the seeds were removed, no amount of watering would regrow the grass. This could happen in a patch of grass or the whole lawn. In this simple analogy the bald patches would be endometrial sclerosis (unstuck Asherman’s), but adhesions would result if two bald patches came into contact with eachother in the uterus.

More recently, the presence of endometrial stem cells at the myometrial junction was reported. These should be able to transform into endometrial stem cells with the right stimulation. The discovery of the existence of endometrial stem cells could explain why in some cases an apparently denuded endometrium (ie.no visible endometrium) will regenerate while in other cases it will not. Or why some women have recurring adhesions while others do not. This would explain why estrogen therapy after hysteroscopic adhesiolysis can in some cases stimulate the regerenation of endometrium while other cases of IUA are recurrent no matter the dose of estrogen therapy and the length of uterine barrier therapy. Endometrium is what keeps the myometrial layer of the uterus from adhering to opposing walls of the uterus. Once again, it is imposible to observe with the naked eye whether stem cells have been removed by curettage or not which would give rise to ‘paradoxical’ outcomes between patients with the ‘same’ severity of injury. In addition, differences in methods for dissecting adhesions can account for differing outcomes. Mechanical methods for adhesion dissection may also be preferable to methods which utilize thermal energy as the latter may damage adjacent tissue.

A modern perspective on an old problem

Authors need to re-evaluate the validity of assumptions, speculations and theories made in publications written when conditions and standards differed greatly to those of today, taking into consideration medical progress.   Although older papers are still important and interesting to read, they need to be interpreted carefully and in the context of limitations in medical technology and critical thinking, lower standards needed for proof and gaps in scientific understanding at the time they were written. Blind repetition of theories cited from papers that were published decades ago without any questioning is simply archaic, unrigorous and unscientific. If there is a case for a genetic or constitutional basis for AS, data from well designed studies would be more persuasive than citing theories which lack any experimental support of any kind. In what seems to be a recurring theme in AS, absence of evidence may not be evidence of absence, but by today’s standards in medicine only presence of evidence is evidence of presence.