mastectomy vs BCT early stage breast, BCT better

[scarbrtj]

IMHO...

The data from more than a quarter million patients from multiple countries shows that BCT offers a statistically significant survival advantage vs mastectomy in early stage breast cancer. Patients should be told that BCT offers a potential survival advantage versus mastectomy. The oncologist is justified in recommending an early stage breast cancer patient who is leaning toward mastectomy to reconsider her decision based on the data (much like one might urge a patient with ER+ Stage I disease with no contraindication to anti-estrogens to reconsider taking an anti-estrogen if refusing to do so).

Discuss!

---

Members don't see this ad.

 

[Chartreuse Wombat]

IMHO...

The data from more than a quarter million patients from multiple countries shows that BCT offers a statistically significant survival advantage vs mastectomy in early stage breast cancer. Patients should be told that BCT offers a potential survival advantage versus mastectomy. The oncologist is justified in recommending an early stage breast cancer patient who is leaning toward mastectomy to reconsider her decision based on the data (much like one might urge a patient with ER+ Stage I disease with no contraindication to anti-estrogens to reconsider taking an anti-estrogen if refusing to do so).

Discuss!

The study design precludes definitive statements about mortality effects of treatment. Correlation is not causation. It is very likely that residual confounding is at play.

 

[scarbrtj]

The study design precludes definitive statements about mortality effects of treatment. Correlation is not causation. It is very likely that residual confounding is at play.

Study design(s) actually, n=130,000, n=132,000, and n=189,000. At this level, an "orgy of data," a different study which you'd like and wouldn't preclude definitive statements about mortality wouldn't convince me much as it'd likely be n=2000 or less (and let's face it, a randomized BCT vs MRM trial is never going to be done again). Keep in mind some other randomized, canonized trials here have shown similar findings. NSABP B-06 5y survival, 82% with mastectomy and 92% with BCT. The NCI, 85% vs 89%. Arriagada and the Goustave-Roussy 15y data said "The relative risks of death and relapse were slightly higher for the mastectomy group than for the tumorectomy group."

So, for real--how much data do we need? A million patients and 20 million patient followup years? I'm asking, for a friend.

 

[Chartreuse Wombat]

Study design(s) actually, n=130,000, n=132,000, and n=189,000. At this level, an "orgy of data," a different study which you'd like and wouldn't preclude definitive statements about mortality wouldn't convince me much as it'd likely be n=2000 or less (and let's face it, a randomized BCT vs MRM trial is never going to be done again). Keep in mind some other randomized, canonized trials here have shown similar findings. NSABP B-06 5y survival, 82% with mastectomy and 92% with BCT. The NCI, 85% vs 89%. Arriagada and the Goustave-Roussy 15y data said "The relative risks of death and relapse were slightly higher for the mastectomy group than for the tumorectomy group."

So, for real--how much data do we need? A million patients and 20 million patient followup years? I'm asking, for a friend.

If the data is not randomized then the sample size could be 1 billion and all that would do is to magnify the biases (and potentially create statistically significant, clinically unimportant results). If you want to make the case that mortality is less with BCT then use the RCTs you quoted and throw out the lower quality studies. Quality of evidence is what is important; not quantity. Epidemiological studies have led us down the wrong path before (see HRT for women).
Again if you are arguing that BCT lowers mortality then stick with the better evidence.

 

[seper]

Just curious, what is the mistake that you're referring to for HRT?

If the data is not randomized then the sample size could be 1 billion and all that would do is to magnify the biases (and potentially create statistically significant, clinically unimportant results). If you want to make the case that mortality is less with BCT then use the RCTs you quoted and throw out the lower quality studies. Quality of evidence is what is important; not quantity. Epidemiological studies have led us down the wrong path before (see HRT for women).
Again if you are arguing that BCT lowers mortality then stick with the better evidence.

 

[Chartreuse Wombat]

Just curious, what is the mistake that you're referring to for HRT?

The weight of epidemiologic evidence suggested that hormone replacement therapy REDUCED the risk of heart attack in postmenopausal women. The Women's Health Initiative a randomized placebo-controlled trial including >100,000 women concluded that estrogen+progestin in fact INCREASED the risk of heart attack (along with increasing risk of stroke, blood clots and breast cancer). The apparent discordance is easily explained by the fact that women who had received HRT included in the epidemiologic studies were DIFFERENT in a number of important characteristics (measured and unmeasured). The point is that epidemiologic studies cannot separate correlation from causation. Only prospective, randomized trials can do that.

 

[scarbrtj]

The weight of epidemiologic evidence suggested that hormone replacement therapy REDUCED the risk of heart attack in postmenopausal women. The Women's Health Initiative a randomized placebo-controlled trial including >100,000 women concluded that estrogen+progestin in fact INCREASED the risk of heart attack (along with increasing risk of stroke, blood clots and breast cancer). The apparent discordance is easily explained by the fact that women who had received HRT included in the epidemiologic studies were DIFFERENT in a number of important characteristics (measured and unmeasured). The point is that epidemiologic studies cannot separate correlation from causation. Only prospective, randomized trials can do that.

You think randomized trials separate correlation from causation? Well OK. I just gave med students a lecture on Werner Bezwoda last week, too. I am at this moment on an airplane attending a large conference on this WHI topic (among others) and have a bulky binder of articles and PowerPoints I'm reading called Mastering the Protocols of Hormone Replacement Therapy. Suffice it to say the Big Lebowski quote "You're out of your element Donnie!" comes to mind. Anyway, back to breast cancer. The data here aren't really conflicting or muddied are they? The birds eye view shows a BCT advantage. The standard dogma is that MRM and BCT are equivalent. If we were aliens coming from another planet and looking at this a-fresh, sans dogma, I think the current thinking would be more open-minded in favor of the data favoring BCT.

 

[Chartreuse Wombat]

You think randomized trials separate correlation from causation? Well OK. I just gave med students a lecture on Werner Bezwoda last week, too. I am at this moment on an airplane attending a large conference on this WHI topic (among others) and have a bulky binder of articles and PowerPoints I'm reading called Mastering the Protocols of Hormone Replacement Therapy. Suffice it to say the Big Lebowski quote "You're out of your element Donnie!" comes to mind. Anyway, back to breast cancer. The data here aren't really conflicting or muddied are they? The birds eye view shows a BCT advantage. The standard dogma is that MRM and BCT are equivalent. If we were aliens coming from another planet and looking at this a-fresh, sans dogma, I think the current thinking would be more open-minded in favor of the data favoring BCT.

You are generalizing again. The point of bringing up WHI was to point out that epidemiologic data frequently is wrong. This is not to say that HRT should not be given to women but -as the FDA and any other credible organization says-It should not be given to lower risks of heart attack. It does provide other benefits such as lessening menopausal symptoms and lowering colon cancer risk but in most women the risks outweigh the benefits. Instead of ad hominem attacks using popular culture films, talking about binders and listing the name of a conference, present a reasoned rebuttal to the central point which is that RCTs are better evidence than observational data.

I hope you made clear to the medical students that Werner Bezwoda was a fraud and his misconduct has nothing to do with the validity of randomized trials. Using his behavior to discredit randomized trials is a disservice to them. You must know that there are cheaters and frauds in every discipline.

I further hope that you did not use the Ionnadis work to discredit randomized trials. If you understand the point that Ionnadis is making and has made for the last decade you know that he is not, by and large, criticizing randomized trials. He is criticizing basic and preclinical research that rely on small samples, multiple hypotheses, ungeneralizable methods and a lack of validation/duplication. A nice short example was published in JAMA earlier this year.

You are painting with much too broad of a brush to use his work to discredit RCTs. Can science be weakened by fraudsters? Of course it can. Are RCTs perfect? Of course not but they remain the best method to get an unbiased estimate of the effects of a treatment. Suggesting that the results of rigorous randomized trials are usually false is to misunderstand the broader point that Ionnadis is making.

 

[scarbrtj]

You are generalizing again. The point of bringing up WHI was to point out that epidemiologic data frequently is wrong. This is not to say that HRT should not be given to women but -as the FDA and any other credible organization says-It should not be given to lower risks of heart attack. It does provide other benefits such as lessening menopausal symptoms and lowering colon cancer risk but in most women the risks outweigh the benefits. Instead of ad hominem attacks using popular culture films, talking about binders and listing the name of a conference, present a reasoned rebuttal to the central point which is that RCTs are better evidence than observational data.

I hope you made clear to the medical students that Werner Bezwoda was a fraud and his misconduct has nothing to do with the validity of randomized trials. Using his behavior to discredit randomized trials is a disservice to them. You must know that there are cheaters and frauds in every discipline.

I further hope that you did not use the Ionnadis work to discredit randomized trials. If you understand the point that Ionnadis is making and has made for the last decade you know that he is not, by and large, criticizing randomized trials. He is criticizing basic and preclinical research that rely on small samples, multiple hypotheses, ungeneralizable methods and a lack of validation/duplication. A nice short example was published in JAMA earlier this year.

You are painting with much too broad of a brush to use his work to discredit RCTs. Can science be weakened by fraudsters? Of course it can. Are RCTs perfect? Of course not but they remain the best method to get an unbiased estimate of the effects of a treatment. Suggesting that the results of rigorous randomized trials are usually false is to misunderstand the broader point that Ionnadis is making.

To summarize, we fart in the general direction* of non-randomized data, no matter the context, quality of analysis, patient numbers, etc? You're generalizing when you say RCTs are better evidence than observational data; there are arguments both ways. I can't be as sure as you!

* movie reference

 

[Chartreuse Wombat]

To summarize, we fart in the general direction* of non-randomized data, no matter the context, quality of analysis, patient numbers, etc? You're generalizing when you say RCTs are better evidence than observational data; there are arguments both ways. I can't be as sure as you!

* movie reference

Potty humor aside I am very confident that a good RCT trumps* observational data anytime. Please be sure to read Ionnadis' paper that you reference (beyond the title).

*trigger warning

 

[seper]

I think you all agree, though, that women should not be getting B/L mastectomy for stage I breast Ca. The more data, the better.

 

[evilbooyaa]

You can make whatever statement but look at the original patient characteristics.

I'm only going to discuss BCT vs Mastectomy alone (as the mastectomy with radiation group will, hopefully we all agree, have higher stage/risk disease on average).

It's in Table 1
I'm not going to copy and paste the whole table, but here's the main points to me:
ER+, 75% in BCT vs 66% in Mastectomy
PR+, 65% vs 55%
Race (other), 8% vs 13% (not only african american patients can be socioeconomically disadvantaged, what about other non-white populations?)
Size < 2cm, 80% vs 64%
Positive LNs, 20% vs 30%

When you're talking about tens of thousands of patients, those 3 variables on percentage basis will be significant.

Sure they controlled for 2 of those, but hormonal positivity? What about margin status (unaccounted for)? Use of estrogen therapy in the HR+ patients between the two groups? Sometimes mastectomy patients may feel like they don't have to take their tamoxifen, since "all the cancer is out"? Her2 status? While I wouldn't expect differences in Her2 status, I also wouldn't have expected a difference in ER positivity, but there was one. What about systemic therapy, or neoadjuvant chemo?

They did a multivariate analysis, based on the variables that they thought to consider. I think that buried in the long list of variables that they didn't consider (or don't have data is from) is where the details lie.

Perhaps you're right, and I will agree that this IS hypothesis generating, but you can't speak from one side of your mouth about how all research is false and then stand behind an epidemiological study and shout 'BCT makes you live longer than mastectomy' from the top of every mountain.

Just my 2 cents.

 

[scarbrtj]

The weight of epidemiologic evidence suggested that hormone replacement therapy REDUCED the risk of heart attack in postmenopausal women. The Women's Health Initiative a randomized placebo-controlled trial including >100,000 women concluded that estrogen+progestin in fact INCREASED the risk of heart attack (along with increasing risk of stroke, blood clots and breast cancer). The apparent discordance is easily explained by the fact that women who had received HRT included in the epidemiologic studies were DIFFERENT in a number of important characteristics (measured and unmeasured). The point is that epidemiologic studies cannot separate correlation from causation. Only prospective, randomized trials can do that.

I would actually love to discuss the WHI and hormone issue more. It's not an easy issue to discuss back and forth online, nor is it even easy to understand, I will admit that. However, a few considerations, of which I find maybe to be the most important to start with... the epidemiological/observational evidence was all positive benefits for estradiol and progesterone... for breast cancer risk, for colon CA risk, for bone health, for heart health. So why in the hell did they use horse estrogens (not pure E2) and progestins (versus pure progesterone) in the WHI when there was negative epidemiological data for those substances? In other words, I can argue that the WHI did *nothing* to negate the epidemiological evidence: it was too different. Second, for the breast CA risk e.g., while we can say that WHI showed "HRT increases breast cancer risk" (with all the caveats of equine estrogens and progestins versus E2 and progesterone), what was the magnitude of that risk? It was an excess 0.8 incidents per 1000 women per year. Substantial? I think not.