wrong result in nested case-control studies
So, in retrospect, we are still looking at why they gave wrong result in nested case-control studies early on. The use of retinol– first of all, think about it. It’s a fat-soluble vitamin. Fat-soluble vitamin means [? what– ?] that you store quite a bit, particularly vitamin A– huge amounts are stored in the liver.
So if you’re measuring diet intake, it is not related to the blood levels. So, if you’re using blood level as a marker for dietary intake, it’s not correct because, irrespective of your diet, if you’re vitamin A level of intake is very low– but still, if you have a lot stored in the liver, blood level will remain normal for a very long period of time. All that for fat-soluble vitamins. So it takes a long, severe, prolonged deficiency for blood levels of vitamin A or other fat-soluble vitamins to be changed.
So that’s not the correct indicator of dietary intake. You are using the wrong indicator. If you’re measuring the association between vitamin A intake, retinol intake, and development of lung cancer, relationship of serum retinol and vitamin A was misunderstood then because there is not much relationship for a long period of time.
These people might have normal serum retinol levels while the vitamin A intake was very low. Or it could have been very high, but it doesn’t show toxicity. Till all the serum-binding vitamin A is full, then it starts attacking the cells. So both very high and low vitamin A intake may not show in serum retinol. So it’s not a good marker– serum retinol– of vitamin intake. This you also need to remember.
Serum retinol levels– I’m repeating– are depressed only with intakes sufficiently low or very high to produce signs of clinical deficiency.
Generally, the range which you see in Western country– will nourished population– serum retinol levels are very minimally related to vitamin A intake because you’ve had so much stored. [? It is ?] you very high or very low. It doesn’t affect blood levels for very long period of time. You have to be very severe in deficiency or very high intake of vitamin A when the blood levels go– change because of the the intake.
So, similarly, large supplements of beta-carotene have no effect on blood levels. It just gets taken out of the blood and stored with the fat, not in the liver but all fat, underlying the tissues.
So the failure to observe an association between serum retinol level and lung cancer tells very little about the relationship between dietary retinol and lung cancer because you were using the wrong biomarker. Blood levels of retinol are not a good marker for vitamin A intake. You need to understand that.
Some other Considerations– the [? consistency ?] was, why did we consistently see, in case-control studies, that there was an association with vitamin A and lung cancer and not in prospective studies? So it makes you think that the– I mean– consistency is one of the main things like in Bradford criteria, which you looked at, that there is association is really true. It’s not spurious. And if the relative risk of our odds ratio is high and it is consistent, that means association is there.
So there was consistent association between carotenoid intake and lung cancer, definitely, not with retinol but with lung cancer and carotenoids. In case-control study, the association was very consistent. And it strongly suggested the relationship is not due to chance.
If you have a large relative risk or odds ratio, measure of association is large and the study is consistent and, of course, if you have dose-response, then you’re pretty sure that the result is– association is not due to chance or some methodologyical problem.
So what was it? So why– what is certain is whether this finding represents a “true causal” effect– that whether beta-carotene is one of the important component causing lung cancer or there is a confounding effect of another dietary or nondietary factors associated with carotene intake. We were not sure in those studies– case-control studies.
They did find a significant association with carotenoids and vitamin A development– beta-carotene. But they did not, really– were certain about the confounding, which could have been due to– if your beta-carotene intake is high, you might be consuming other fruits and vegetables, which were doing the thing– decreasing the risk– or some other nondietary factor which is associated with intake of beta-carotene. That was not taken into account. So there could have been confounding, which was not a residual confounding or confounding, which was not taken into account.
So it is very hard to exclude possibility that if your fruit and vegetable intake is high, you may have a lot of other things. And you measure just beta-carotene and you say, beta-carotene decreases the risk of lung cancer. But the food which provided you a lot of beta-carotene had many other things like indole oil or sulfur-containing compounds or other biochemicals or fiber or low fat because you eating fruit and– hundreds of different other confounding factors, which could be truly protective. And they were varying with the beta-carotene levels. So there was confounding because of the strong correlation between beta-carotene and other protective factors. And that may have been decreased in [INAUDIBLE] of lung cancer.
So what you can summarize is that available data best interpreted is that there is strong support for what you can say, which is consistent from very long period of time and it’s still true, that the protective effect is of fruits and vegetables against lung cancer.
We’re not sure what’s in fruit and vegetable. Maybe beta-carotene and alpha-carotenoids and many other carotenoids working with some other things. Or it may not be carotenoids at all. It may be that the [INAUDIBLE] with some other things in fruit and vegetable beta-carotene and the carotenoid which is decreasing the risk.
So sometimes, that is a confounding. You need to replace– [? correct ?] for that. When the two things vary similarly though, one of them is the real cause and other just varies with it. So you have to control for it. But we do not know what are the other things that should be controlled for in fruits and vegetables. But we do know, for sure, that’s easier thing to reduce the risk of lung cancer and many other chronic diseases is to– which is a bonus that increases your intake of fruits and vegetables.
If you increase that, risk of many chronic diseases not only lung cancer goes down. And also, it helps to regulate your blood glucose level. It gives you more fiber probably, which decreases– which helps to regulate, probably, blood glucose levels.
Many chronic diseases, including obesity, goes down if your intake of fruit and vegetable is high because you can eat only so much. So if you substitute fruits– you eat a lot fruits and vegetables, you can eat only so much and you get full. So you will eat less of animal products, animal fats. Maybe they are the ones which increase the risk. So your total calorie intake may go down and especially if your whole fruits and vegetables– you’re eating more fiber.
So there are lots of benefits of increasing the fruits and vegetables, one of them being, consistently– that it decreases the risk of lung cancer and many other cancers like stomach cancer, throat cancer, and even the breast cancer– so the strong evidence that fruits and vegetables definitely decrease the risk.
But most prospective studies, later on, and the pooled prospective studies did not show much association with carotenoids and nothing with retinol. Also, case-control studies did show [? somewhat ?] consistently. But then, the other ones, which were– nested case-control were reversed if the follow-up was long.
So it seems that there’s a lot of confusion and not really much association between either case-control or prospective, even large pooled studies. Some of large pooled studys shows that intake [? of ?] large supplementation with beta-carotene and lutein actually increase the risk.
So, whenever you have repeatedly done case-control studies, you’ve done a lot of prospective studiess, and the question is still not clear, though you have a pretty good idea of which way it’s going, is so many confounders, which you cannot control for. The [? idealists ?] thinks [? that ?] which comes as close to experimentation with humans is to assign an exposure and then see where there’s a difference. Those who are exposed were very similar to unexposed in everything except the exposure which you are interested in. So that means randomized clinical trial.
So then, later on, in late -90s, early 2000s, many different randomized clinical trials were done. And that pretty much answered the question, so– the highest hierarchy of study design randomized clinical trials are at the top. You cannot do clinical trials from anything because it means assigning an exposure, which may be unethical sometimes.
It is done only after you have a lot of results, which are either confusing or point to one way from case-control and for prospective study, then you do the randomized clinical trial if you can do it ethically. And it seems to clear up the confusion.
So we will next time discuss the randomized clinical trial, which many of them were done late ’90s, early 2000s, to see if there is an inverse association between various types of carotenoids, including beta-carotene and risk of lung cancer. And we’ll do that next time. Thank you.