In this paper, it says that "Although the prevalence of pineal cysts in autopsy series has been reported as being between 25% and 40%, MR studies have documented their frequency to range between 1.5% and 10.8%." I think the MR studies is not representative of general population, since the people who would go to have an head MR scan usually have some related symptom. What about the result from autopsy?

Are the people who are autopsied representative of general population? I think that it is not the case since it is usually those that die of unknown causes, e.g. those that die of traffic accident would not be autopsied, and especially not brain autopsied.

  • If you take all autopsies from whatever reason and all MRIs from whatever reason, autopsies would be more representative, because they would include broader list of background conditions. Note that when autopsy is done, it's usually done completely, not just as "brain autopsy" or so.
    – Jan
    Jul 1 '20 at 14:19
  • What do you mean by "all autopsies from whatever reason"? You can not force someone's relative into allowing an autopsy. You can not force MRI on someone either. Would you mind if you cite a source for "it's usually done completely, not just as "brain autopsy" or so"?
    – Aqqqq
    Jul 1 '20 at 14:25
  • Autopsies are done when required to find the cause of death in uncertain cases and are sometimes required by law. The results of autopsies are then available for various researches. Autopsy can reveal different conditions in different parts of the body. MRI of the head reveals only the conditions in the head and is done only when a condition in the head is suspected. So, I would say autopsies are more representative. Autopsies are typically always done completely and in the same order. I'm just commenting, not answering. You can read about autopsy elsewhere.
    – Jan
    Jul 1 '20 at 14:34
  • With other words, pineal glands from autopsies are from more different sources (very different conditions) than those from head MRIs (head conditions only), so more representative of general population.
    – Jan
    Jul 1 '20 at 14:38
  • But is it sufficiently representative of general population? Would the occurance of the cyst in autopsy be higher than general population since the cyst can be a cause of sudden death?
    – Aqqqq
    Jul 1 '20 at 14:48

Every sample is biased.

The question is whether these samples are meaningfully biased in a way that would like affect these results.

Yes, those getting a head MRI likely have some symptoms related to their head or brain area, but the vast majority of those will have nothing to do with pineal cysts or tumors; in addition, some of the work in the paper you cite was on healthy volunteers. Healthy volunteers are often over-represented by college-aged students, which likely don't overlap as much with the medical patients.

Similarly, those who require autopsy are not the same of the general population, but I can't think of any reason that the population getting autopsies would have a strong relationship to the presence of pineal cysts.

Pu et al, the paper you cite, refer to differences with age and gender, and it's likely that both MRI- and autopsy-receiving cases differ on those groupings, but not to an extent that would produce a 20-fold difference in prevalence.

So then why such a big difference in autopsy versus imaging results?

I presume this question is what really led you to asking about the autopsy studies. Both the paper you link to, and a related one (Al-Holou, et al 2011) provide some quick clues. Quoting from Al-Holou, et al:

It is not surprising that autopsy studies, because they include cysts that may be too small to be detected on MR imaging, have reported consistently higher prevalence rates for pineal cysts, ranging between 21% and 41%.25,65 In contrast to imaging studies, autopsy studies have included small or even microscopic cysts. For example, in the series reported by Hasegawa et al.,25 45% of all pineal cysts were 2 mm or less in maximum diameter. This detection variability may be a factor even between different estimates of prevalence on MR imaging.

The paper you refer to, Pu et al 2007, specifically used a finer-detail scan:

A possible explanation for the difference between the prevalence of the pineal cysts demonstrated in our study and that in other MR imaging studies is that we used high-resolution MR imaging with a 1.0 mm resolution on the 3D T1WI and an in-plane resolution of 1.0 mm with an axial resolution of 2.0 mm on the T2WI. An in-plane resolution of 1.0mm and an axial resolution of 3.0mm were used in the prior MR imaging study of healthy volunteers.17

If 45% of cysts from autopsies are themselves smaller than 2mm, it isn't reasonable to expect that most of those will be found on MRI scans, especially those with 3mm spacing.

The autopsy numbers are certainly more accurate from a cyst-counting perspective, but as the Pu et al paper states:

Most cysts are asymptomatic, with diameters ranging from 2 to 15 mm.5-8 When present, however, symptoms are usually noted in patients with cysts larger than 15 mm in diameter.9-10

If only cysts larger than 15mm are commonly clinically relevant, all we learn from counting the smaller cysts is in confirming that small cysts are common and not normally associated with symptoms.

Probably the best estimate of general prevalence would come from a pre-registered sample of patients whose brains would be donated specifically for pineal histology post-mortem. I can't see a lot of support or funding for such a study, though, given the lack of clinical relevance.

Al-Holou, W. N., Terman, S. W., Kilburg, C., Garton, H. J., Muraszko, K. M., Chandler, W. F., ... & Maher, C. O. (2011). Prevalence and natural history of pineal cysts in adults. Journal of neurosurgery, 115(6), 1106-1114.

Pu, Y., Mahankali, S., Hou, J., Li, J., Lancaster, J. L., Gao, J. H., ... & Fox, P. T. (2007). High prevalence of pineal cysts in healthy adults demonstrated by high-resolution, noncontrast brain MR imaging. American journal of neuroradiology, 28(9), 1706-1709.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.