The buspirone challenge test clearly distinguishes ME/CFS patients from healthy controls: why is it not being developed and deployed?

[V.R.T.]

Putting on my amateur musing hat again: I think a lot could potentially be done with a single patient, and maybe it could be done much faster than assembling a big new study. Some of the studies were finding ME/CFS prolactin responses far above what healthy individuals showed. When lab findings are nowhere near what you expect for healthy people, you can learn a lot without needing the statistical power of large groups. The hope is that whatever patient would volunteer would be one of those with a highly abnormal response, and that ideally other test findings would be very abnormal as well, like maybe prolactin response to TRH.

It might just take a motivated clinician-researcher and a patient who is willing to be a bit of a guinea pig (and a dollop of ethics approval and a spoonful of cash). The case study on the patient would provide motivation to replicate whatever is found in an actual case-control study of ME/CFS.

This sounds very much like something Fluge and Mella could conceivably take on. Or someone similar. I'd say Scheibenbogen but I think her team has a tendency to put too much spin on their results sadly. But if we think within the field for this initial case study maybe we can find someone fairly quickly.

 

[forestglip]

Could an imbalance of the short and long forms of prolactin receptors mean that some cells are going into apoptosis or be otherwise misbehaving in some way? Perhaps they’re responding badly to perfectly normal signals?

Could the increase that we see in response to Buspirone be a result of the body over compensating to this and trying to get a signal to somewhere and it's just not being heard so it shouts louder? Maybe some subset of cells up or down regulating in response to this imbalance of receptors elsewhere?

But wouldn't this lead to high prolactin all the time, if it's compensating for not getting a signal across?

 

[hotblack]

But wouldn't this lead to high prolactin all the time, if it's compensating for not getting a signal across?

Yes quite possibly. Which is one of the reasons I was interested in some mechanism that wasn’t a simple up or down regulation but whereby you had a noisy and spiky system.

Just throwing ideas around. And even with this prolactin receptor imbalance idea it would be in a specific cell population or even localised rather than necessarily something happening at the hypothalamus and pituitary gland. So perhaps therefore not leading to a constant elevation. All a bit vague and possibly unrelated sorry. I’ve just become intrigued by some of the mechanisms and process around these receptors and their uses and effects in different cells. And wondering about if or how it could be linked.

 

[forestglip]

I think the Sharpe 1996 (Cowen group) study had a pretty good view on the problem saying it was likely the dopamine instead of serotonin pathway that was abnormal. They wrote:

Our data question whether the enhancement of buspirone-induced prolactin release in CFS is a consequence of increased sensitivity of post-synaptic 5-HT,, receptors. It is possible that the increased prolactin response to buspirone in CFS could reflect changes in dopamine function.

Increased prolactin response to buspirone in chronic fatigue syndrome - PubMed

But even this seems to be overly focused on specifically the receptors that the drugs bind to, as if an abnormality in those receptors is the only way one would see an increased response. They don't seem to be considering that it could be anything beyond those receptors. Even if buspirone is causing it through the dopamine receptors, could the abnormality not still be downstream of that, within the lactotrophs themselves?

And they went right back to focusing on serotonin after seeing the d-fenfluramine studies (for example, in Sharpe's 1998 book).

Here is an example of not being focused only on the receptors being probed, from a study of primary testicular failure: Exaggerated Prolactin Response to Thyrotropin-Releasing Hormone and Metoclopramide in Primary Testicular Failure (Spitz et al., 1980)

They saw an increased prolactin response to the dopamine antagonist, metaclopromide (and also to TRH). Notably, they did not at all speculate that this might relate to changes in dopamine function. They know that estrogens cause such a situation to happen, so they speculate about how estrogens may be involved, even though there were no direct signs of estrogen involvement in the study.

The fact that castrates have PRL responses indistinguishable from normal implies that the exaggerated profile in testicular failure is related
to the secretion of a testicular secretory product. The testis can be defective as regards spermatogenesis, but its steroidogenic potential must remain intact.

Although E2 [estradiol] is a well-known stimulator of PRL,22, 23 mean E2 levels were increased only in group III subjects who had the least exaggerated PRL response. Moreover, in contrast to the gonadotropins there was no direct relationship between E2 and the PRL response.

Although in previous studies 7 we found a correlation between the E2:T [estradiol:testosterone] ratio (indicative of a relative increase in E2) and the peak PRL response, such a correlation was not confirmed in this large number of patients. Again, the highest E2:T ratio occurred in group III. Although E1 [estrone] was increased, this also did not correlate with the PRL response. Estriol is also unlikely to be incriminated directly. 24

The fact that the estrogen antagonist clomiphene citrate has the ability to decrease the exaggerated PRL response nevertheless suggests that estrogens do play a role in this phenomenon.

(Line breaks added)

 

[forestglip]

Here is an example of not being focused only on the receptors being probed, from a study of primary testicular failure: Exaggerated Prolactin Response to Thyrotropin-Releasing Hormone and Metoclopramide in Primary Testicular Failure (Spitz et al., 1980)

They saw an increased prolactin response to the dopamine antagonist, metaclopromide (and also to TRH). Notably, they did not at all speculate that this might relate to changes in dopamine function. They know that estrogens cause such a situation to happen, so they speculate about how estrogens may be involved, even though there were no direct signs of estrogen involvement in the study.

In the same vein, and from the same research group, this study showed an increased prolactin response to TRH in primary ovarian failure:

Increased prolactin response to thyrotropin-releasing hormone in primary ovarian failure (Hochner-Celnikier et al., 1982)

I thought it was worth mentioning because they suggest an interesting possibility for how estrogens might cause increased prolactin response, even when the study cohort showed low, not high, blood estrogen levels. They suggest that there might be increased conversion of androgens to estrogens specifically within the hypothalamus. In support of this was the observation that when rabbits are castrated, they show increased production of estrogens in the hypothalamus, and this might relate to the ovarian failure seen in the study cohort.

It is well known that estradiol increases PRL responsiveness. [20-23] However both E2 and E1 levels were markedly reduced in the patient group. Nevertheless, the postmenopausal ovary does secrete estrogens [24, 25]; Gronroos et al [26] found higher mean serum levels of E1,
E2, and total estrogens in ovarian effluent than in blood. Another important source of estrogens in the postmenopausal woman is the peripheral conversion of androstenedione to E1, [27] which correlates with body weight. [28] However, none of the authors’ patients was obese.

Naftolin and Ryan [29, 30] demonstrated in vivo and in vitro conversion of androstenedione and testosterone to E1 and E2 in the anterior hypothalamus from humans, rhesus monkeys, rabbits, rats, and mice. The aromatizing activity was increased in castrated rabbits. [31] This phenomenon might be operating in the subjects described without demonstrable effect on peripheral estrogen levels. As production of androstenedione is in milligram amounts, even a small percent conversion could result in significant contribution to estrogens.

The catechol estrogen 2-hydroxyestrone, which is formed in the pituitary gland, [32] is another potential stimulus of PRL secretion. [33] It should also be mentioned that in addition to stimulating PRL secretion, estrogen also sensitizes the lactotrope to the inhibitory effect of dopamine on PRL release. [34, 35] In a situation of estrogen deficiency, such as in the present study, there is presumably decreased sensitization of the lactotrope to dopamine. This could also be a factor that explains the exaggerated PRL response. However, it is also likely that factors other than estrogens are responsible for the exaggerated PRL response to TRH in primary ovarian failure. Further studies to elucidate this are now in progress in the authors' laboratory.

Spoiler: References

20. Yen SSC, Ehara Y, Siler JM: Augmentation of prolactin secretion by estrogen in hypogonadal women. J Clin Invest 53:652, 1974

21. Buckman MT, Peake GT: Estrogen potentiation of phenothiazine induced prolactin secretion in men. J Clin Endocrinol Metab 37:977, 1973

22. Carlson HE, Jacobs LS, Daughaday WH: Growth hormone thyrotropin and prolactin responses to thyrotropin-releasing hormone following diethylstilbestrol pretreatment. J Clin Endocrinol Metab 37:488, 1974

23, Rakoff JS, Siler TM, Sinha YN, et al: Prolactin and growth hormone release in response to sequential stimulation by argi- nine and synthetic TRH. J Clin Endocrinol Metab 37:641, 1973

24. Mattingly RF, Huang WY: Steroidogenesis of the menopausal and post menopausal ovary. Am J Obstet Gynecol 103:679, 1969

25. Struthers RA: Post menopausal estrogen production. Br Med J 1:1331, 1956

26. Gronroos M, Klemi P, Salmi T, et al: Ovarian production of estrogens in post menopausal women. Int] Gynecol Obstet 18:93, 1980

27. Siiteri PK, MacDonald PC: Role of extraglandular estrogen in human endocrinology, Handbook of Physiology, Section 7, Endocrinology. Edited by SR Geigen, EB Astwood, RO Greep. Washington, DC, American Physiology Society, 1973

28. Siiteri PK: Estrogens in the postmenopause. Front Horm Res 3:40, 1975

29. Naftolin F, Ryan KJ: The metabolism of androgens in central neuroendocrine lissues, J Steroid Biochem 6:993, 1975

30. Ryan KJ, Naftolin F, Reddy V, et al: Estrogen formation in the brain. Am J Obstet Gynecol 114:454, 1972

31. Vangala V, Reddy R, Naftolin F, et al: Aromatization in the central nervous system of rabbits: Effect of castration and hormone treatment. Endocrinology 92:589, 1973

32. Fishman J, Naftolin F, Davis IJ, et al: Catechol estrogen formation by the human fetal brain and pituitary. J Clin Endocrinol Metab 42:177, 1976

33. Adashi EY, Casper RF, Fishman J, et al: Stimulatory effect of 2- hydroxyestradiol on prolactin release in hypogonadal women. J Clin Endocrinol Metab 51:413, 1980

34. Neill JD, Frawley L, Plotsky PM, et al: Dopamine in hypophysial stalk blood of the rhesus monkey and its role in regulating prolactin secretion. Endocrinology 108:48, 1981

35. Judd Sj, Rigg LA, Yen SSC: The effects of ovariectomy and estrogen treatment on the dopamine inhibition of gonadotropin and prolactin release. J Clin Endocrinol Metab 49:182, 1979

Edit: However, I'm not sure how to square this with what they said in the quote from their other paper, posted a couple posts up, where they said the prolactin response in castrates is indistinguishable from normal. If increased conversion to estrogens in the hypothalamus is responsible for increased prolactin response in ovarian failure, and if castrated rabbits show this increased conversion, why don't castrated animals show increased prolactin response? This is what they said has been seen in castrates, from the other paper:

The mechanism of the increased PRL response in severe oligospermia and azoospermia is unknown. In individual patients with testicular failure and in control subjects, the PRL response was directly related to basal and peak gonadotropin responses to LHRH. However, this cannot be the major factor since male castrates with markedly elevated gonadotropin levels have intact PRL responses. 19 Furthermore, in the castrated rat, there is a decreased PRL response to both TRH and MET.20

 

[obeat]

For a case study would we not need one pre menopause woman and one postmenopausal woman.?

 

[forestglip]

For a case study would we not need one pre menopause woman and one postmenopausal woman.?

The more people that are studied, the more we can learn, of course. But I think even just one individual could provide useful insights, and might move things along faster, as long as that person has a substantially increased prolactin response to buspirone.

Though if a clinician went to their ME/CFS patients and tested each one to find a candidate with such an increased response, and they had to test 5 different patients to find one, then I'd be concerned about why the first 4 didn't have an increased response. But if the first patient they check shows this pattern, then I'd be fairly comfortable with conclusions based on that one individual.

 

[ME/CFS Science Blog]

I've been extracting the data from the graphs in the buspirone studies, it looks something like this. For Bakheit 1992 I took males only, Sharpe 1996 is also males only, the other studies have both sexes combined.

EDIT: there might be an error in the Sharpe 1996 original data plot, see the discussion in the comments below.

 

[forestglip]

Note that it seems like for Sharpe 1996, the minutes on the x-axis in their paper might have been doubled, as ScoutB suggested.

 

[ME/CFS Science Blog]

Note that it seems like for Sharpe 1996, the minutes on the x-axis in their paper might have been doubled, as ScoutB suggested.

Thanks, I've updated the plot accordingly.

 

[forestglip]

Thanks, I've updated the plot accordingly.

Sorry, more about this.

Good thinking. Trying to reconcile this and I think the Sharpe study may have mislabelled the x-axis of their graph. Here are their methods:

After a 30-min rest period, venous samples were removed for baseline hormone estimations. Following this, buspirone (0.5 mg/kg up to a maximum of 45 mg) was administered at time ‘0’ and further venous samples removed at 15-min intervals until + 120 min. A final sample was taken at + 150 min. Subjects rested semi-supine throughout the blood sampling period and were not allowed to sleep. They rated themselves on IO cm visual analogue scales (VAS) for ‘nausea’ and ‘lightheaded’ at baseline and at 30-min intervals throughout the period of buspirone testing.

So you'd expect the graph to show data points every 15 minutes up to 120 or 150 minutes. Instead the graph (see post #191) shows data points every 30 minutes up to 240 minutes. If you assume they made a mistake and each notch is 15 minutes instead of 30 it becomes a graph up to 120 minutes, with controls peaking at 90 minutes as in your examples. (And maybe they dropped the last 150 minute reading because it made the graph harder to read without adding much info??)

I wonder if the mistake was actually in describing the intervals as 15 minutes in the text, while the plot with 30 min intervals is correct. 15 minute intervals would be very frequent. I could certainly be forgetting a study, but I don't recall seeing any that measured prolactin more frequently than 30 minutes in other health conditions. I just checked 5 prolactin response studies at random, and they all used 30, 45, or 60 minute intervals.

Plus, the 240 minute time span for testing exactly matches two other ME/CFS studies if taking their plot at face value.

It does seem to lead to the peak in controls being twice as delayed as is usually seen, so that's a point against trusting the plot, though.

So I'm not sure, but I might lean towards just assuming the plot is the accurate one.

 

[ScoutB]

I've been extracting the data from the graphs in the buspirone studies, it looks something like this.

Super nice. We could add these or something similar (forestglip has made a few as well) in the google doc and maybe the set of info we're sharing around. Seeing them at once gets the message across pretty clearly.

 

[ScoutB]

So I'm not sure, but I might lean towards just assuming the plot is the accurate one.

Two other things we could check are if the 30 minute intervals which are also on the growth hormone plot make sense (is the peak consistent with other studies?) and if anyone knows how to calculate the time effect thing (which resulted in, for example, a value of 396 +- 140 ng x min / ml for the control curve) I guess we could see which interval it's consistent with.

Can't prove it either way, but maybe if everything but the x-axis is consistent that'd give us some confidence one way, and if everything but the "15 minutes" in the methods is consistent then the other way might be safer.

Frustrating to effectively lose information like this.

 

[ME/CFS Science Blog]

I might lean towards just assuming the plot is the accurate one.

Yes that might be the more conservative option. The plot does show 15-minute intervals before the administration of buspirone. And as ScoutB says, the growth hormone plot also shows the same intervals up unto 240 so perhaps it's simply a bad description in the methods section.

 

[forestglip]

He reports the patient selection in such an ambiguous way I can’t tell how many male vs female patients he had (you'd think "five males and twenty five females were studied" would do it, but no, he goes on to say he took a mystery subset of 25 of those for the actual test).

For Richardson 1995, I think they excluded 5 patients. No indication of which sex the excluded patients were.

Taking a look at Richardson 1995 again, I noticed that I previously missed a sentence that suggests that he did actually test all 30 patients.

From the methods, I thought this meant that only 25 patients were used in the study - those with a positive VP1 test:

Five males and twenty-five females were studied, all with an extended history of muscle fatigue associated with cognitive disorders. [...] Twenty-five of those with a positive VP1 test, that is of those with values for binding of the monoclonal antibody more than two standard deviations above that of normal controls, were used in this study.

But the discussion includes this sentence saying all 30 patients had a prolactin ratio. So I think the earlier sentence was just saying that out of the 30 patients, 25 had a positive VP1 test.

In the patients, the mean figure for the prolactin ratio is three times higher that that in the controls, a highly significant difference. A ratio of 2.5 and upwards encompasses 26 out of 30 patients and a ratio less than this defines a similar proportion of controls, assuring a reliability for the distinction of 87%.

 

[Axel]

But even this seems to be overly focused on specifically the receptors that the drugs bind to, as if an abnormality in those receptors is the only way one would see an increased response. They don't seem to be considering that it could be anything beyond those receptors

Do studies examines receptors as elements capable of adapting to a given (abnormal) situation? In reality, merely referring to the sensitivity of receptors provides no insight into the upstream and downstream mechanisms. This focus also reflects the pharmaceutical industry’s ability to produce molecules that block receptors or inhibit the reuptake of neurotransmitters.
--
Assuming that lactotrophic cells is more active but basal level normal, I suppose that either the receptors have become more numerous or more sensitive, enabling them to block increased prolactin production with a normal amount of dopamine.

Or perhaps the amount of dopamine is regulated upwards further up in the circuit.

But in a way, it doesn’t matter. If there is an excess of pression, it would be more interesting to identify either the triggering event (which has now ended and been compensated for) or the element that would be still active.

 

[ScoutB]

I noticed that I previously missed a sentence that suggests that he did actually test all 30 patients.

True.

I get the feeling he (and maybe others in the field?) thought the enterovirus test was important for cohort selection at the time.

Imo, that first line you quoted still reads to me like 'the study' is about the 25 patients. In the abstract he also talks about the study as if it's only on the VP1 positive group:

The buspirone-prolactin reponse was studied in a subgroup of patients with CFS/ME and evidence of persistent enteroviroal infection, as shown by the repeated detection of the group-specific protein of enteroviruses, VP1, in the blood.

But I think you're probably right that he tested all 30, and maybe because people were keen on the enterovirus idea he described 'the study' focusing on the VP1+ 25 as if they were all that existed.

Or maybe he's calling it a study on enterovirus ME/CFS patients because he did recruitment in a special way to get so many (I assume we don't think 25/30 ME/CFS patients today have an easily diagnosed enterovirus infection?)

More focus on the enterovirus patients at the end:

The studies here show that in CFS/ME patients with persistent enteroviral infection there are significant changes in hypothalamic functions. In view of more recent studies which show changes in hypothalamic blood flow in these patients, the time would seem ripe to determine what the somatic signals are which can lead a persistent virus infection to cause such changes in the brain. [...] It is also important to extend these observations to other subgroups of CFS/ME patients.

And btw I'd like to know more about those hypothalamic blood flow studies...

 

[forestglip]

Am re-reading the studies anyway, so here's an example of the table I had in mind:

Study	Patients - controls	Stimulant	Effect on prolactin	Notes
Bakheit 1992 (Behan group)	15 post-viral fatigue syndrome patients 40% female 13 healthy controls 13 depressed controls	60 mg buspirone	Similar baseline but prolactin levels rose 4.9 (males) – 8.2 (females) fold in patients, compared to 1.7 and 3.5 fold in depressed controls and 2.4 and 3.8 fold in healthy controls	Peak was after 1 hour of 3 hours tested Females tested during luteal phase “The buspirone caused excessive fatigue, lightheadedness, and nausea in patients but not in controls”
Clear 1995 (Wessely group)	10 CFS patients Holmes and Oxford criteria 40% female 20 healthy controls 10 depressed controls	30 mg d-fenfluramine	1.5 fold increase in patients compared to less than 1.16 increase in controls	Peak after 3-4 hours, 5 hours tested. Females tested during follicular phase Cortisol showed no significant group changes
Bearn 1995 (Wessely group)	9 CFS patients Oxford criteria 44% female 10 controls	30 mg d-fenfluramine	No significant group difference in the prolactin response	Cortisol showed no group difference; ACTH response was greater in patients
Bearn 1995 (Wessely group)	9 CFS patients Oxford criteria 44% female 8 controls	Insulin tolerance test	Lower response in patients: Almost 7 fold increase in controls, only 4 fold increase in patients	Growth hormone also showed slightly lower response in patients, cortisol or ACTH showed no group difference
Yatham 1995	11 CFS patients Fukuda criteria 72% female 11 healthy controls	60 mg dl-fenfluramine	No group difference	5 hours tested, cortisol also showed no difference
John Richardson 1995	25 CFS/ME patients ca. 83% female Fukuda and Oxford criteria 25 controls (family members)	50 mg buspirone	Similar baseline but prolactin levels rose almost 7 fold in patients, compared to 2 fold in controls	Single author, a GP not an academic “Nausea in patients as a response to buspirone was very marked in comparison to controls, and in most cases predicted the outcome of the test.”
Sharpe 1996 (Cowen group)	11 CFS patients Oxford criteria 0% female 11 controls (hospital and university staff)	0.5 mg/kg 45mg max buspirone	Prolactin levels increased ca. 2.6 fold in patients compared to ca 1.7 fold in controls	Peak at 1.5 hour, 4 hours tested Growth hormone showed no significant changes Patients also had more nausea in response to buspirone but this did not correlate with prolactine response Plasma levels of buspirone and its major metabolite I-(2-pyrimidinyl)piperazine (I-PP) were not significantly different between groups. They found “excessive variance in buspirone-induced prolactin release in both female CFS subjects and controls.”
Behan 1996	10 CFS patients with chronic exposure to organophosphate (OP) insecticides. 0% female 30 controls from previous study (50% female)	60 mg buspirone	Ca. 5 fold increase in patients compared to ca 2 fold increase in controls.	Single authors study and Behan was later found to have made errors, e.g. stating the controls were 10 males. Controls were matched for sex Patients had lower growth hormone response to pyridostigmine and dexamethasone
Majeed 1996 (Thesis)	30 CFS patients Fukuda criteria 50% female 30 healthy controls	60 mg buspirone	Similar baseline but ca. 2.7 fold increase in patients compared to ca. 1.8 fold increase in controls.	Tested for 4 hours “none of our subjects reported any feelings of nausea.” Growth hormone release after bromocriptine showed no difference (page 145 of thesis). Results never published
Sharpe 1997 (Cowen group)	10 CFS patients Oxford + neurasthenia criteria 0% female 10 controls	30 mg d-fenfluramine	Ca. 1.6 fold increase in patients compared to stable levels in controls	Levels first dropped after 1 hour then rose with the biggest difference appeared after 4 hours
Richardson & Da Costa 1998	39 CFS patients Fukuda and oxford criteria 56% female	50 mg Buspirone	Prolactin rose ca. 4.5 fold in males and ca. 7.8 fold in females.	No control group. One female patient received only 20 mg because she had epilepsy and showed less prolactin increase
Ottenweller 2001 (Natelson group)	20 CFS patients Holmes and Fukuda criteria 100% female 14 controls	Maximal exercise test on treadmill	Lower response in patients: ca. 2 fold increase in controls versus only ca. 1.2 fold in patients	Adrenocorticotropin, epinephrine, thyrotropin responses were lower, growth hormone higher and norepinephrine the same as controls
Racciatti 2001	14 CFS patients 5 after toxic exposure, 4 after EBV, and 5 with comorbid depression 71% female No controls	Buspirone	“…an abnormal increase of prolactine levels followed the buspirone challenge test”	No figures or data reported Dosage unclear No control group.
Sharma 2001	1 female CFS patient	30 mg Buspirone	10 fold increase in prolactin during illness but stable response after successful graded exercise program	No control group
Vasallo 2001 (Cowen group)	20 CFS patients (4 male, 16 female) 21 controls	m-chlorophenylpiperazine (mCPP)	No group difference: prolactin doubled in both patient and controls,	Patients also received a placebo which didn’t raise prolactin
Weaver 2010 (Natelson group)	22 CFS only 68% female 11 CFS + fibromyalgia 72% female 16 controls 62% female	120 mg of l-tryptophan per kg	“Women with CFS alone, but not CFS + FM, showed upregulated plasma PRL responses compared with Controls” Ca. 2 fold increase in female CFS only versus 2 fold in female controls. No difference in males.

I used your table as a basis to make one that I hope to include in the review:

Study	Participants	Probe	Prolactin response*	Notes
Buspirone​
Bakheit, 1992 Bakheit et al., 1992 (Behan group)	6F, 9M PVFS 6F, 7M depression 6F, 7M healthy	Buspirone (60 mg, orally)	PVFS: 8.2 (females) and 5.0 (males) Depression: 3.6 (females) and 2.1 (males) Healthy: 3.9 (females) and 2.5 (males)	Females tested during luteal phase. “The buspirone caused excessive fatigue, lightheadedness, and nausea in patients but not in controls”
Richardson, 1995	25F, 5M CFS/ME (Oxford) 30 healthy (Details of sex not reported)	Buspirone (50 mg, orally)	CFS/ME: 4.9 Healthy: 1.8	Females tested during luteal phase. Patients experienced more nausea after buspirone compared to controls.
Sharpe et al., 1996 (Cowen group)	11M CFS (Oxford) 11M healthy	Buspirone (0.5 mg/kg up to 45 mg, orally)	CFS: 2.9 Healthy: 2.4	Growth hormone response was not significantly different between groups. Patients had more nausea in response to buspirone but this did not correlate with prolactin response. Plasma levels of buspirone and its major metabolite 1-(2-pyrimidinyl)piperazine (1-PP) were not significantly different between groups.
Behan, 1996	10M CFS with previous exposure to organophosphates. 15F, 15M healthy from previous study	Buspirone (60 mg, orally)	CFS: 4.8 Healthy: 1.9	The paper was later found to include an error: the control group was incorrectly described as 10 males. Patients also had increased GH response to pyridostigmine and decreased GH response to dexamethasone.
Majeed, 1996 (Thesis, supervised by Peter O. Behan)	15F, 15M CFS (Fukuda) 15F, 15M healthy	Buspirone (60 mg, orally)	CFS: 2.8 Healthy: 2.0	Females tested during follicular phase. All participants had been medication-free for at least 3 months. No participants reported nausea after buspirone. Normal GH responses to bromocriptine and baclofen, larger GH response to pyridostigmine, smaller GH responses to desipramine and dexamethasone, and smaller ACTH response to ipsapirone.
Racciatti et al., 2001	3F, 2M CFS after toxic exposure 3F, 1M CFS after EBV 4F, 1M CFS with depression No healthy controls	Buspirone	“[…] an abnormal increase of prolactine levels followed the buspirone challenge test […] in all the three subgroups of patients.”	Prolactin values and buspirone dosage not included in paper.
Sharma et al., 2001	1F CFS (Oxford)	Buspirone (30 mg, orally)	CFS: 10.9 After symptom improvement following graded exercise program: 12 months: 2.1, 18 months: 1.2	Case study.
Other probes​
Cleare et al., 1995 (Wessely group)	4F, 6M CFS (Holmes and Oxford) 4F, 6M Depression 8F, 12M healthy	D-fenfluramine (30 mg, orally)	CFS: 1.5 Depression: 0.9 Healthy: 1.2	Females tested during follicular phase. Cortisol response did not differ from healthy group. When analyzing full cohort, baseline cortisol inversely correlated with prolactin response.
Bearn 1995 (Wessely group)	4F, 5M CFS (Oxford and all but one fulfilled Holmes) 6F, 4M Healthy	D-fenfluramine (30 mg, orally)	CFS: 1.5 Healthy: 1.7	Females tested on days 3-5 of their menstrual cycle. ACTH response was greater in patients. Cortisol response showed no group difference.
Bearn 1995 (Wessely group)	4F, 5M CFS (Oxford and all but one fulfilled Holmes) 6F, 2M Healthy	Insulin (0.1 or 0.15 u/kg)	CFS: 5.1 Healthy: 6.7	Females tested on days 3-5 of their menstrual cycle. Growth hormone response trending lower in patients. Cortisol and ACTH responses were not significantly different.
Yatham et al., 1995	8F, 3M CFS (Holmes) 8F, 3M Healthy	DL-fenfluramine (60 mg, orally)	CFS: 1.7 Healthy: 2.5	Cortisol response did not significantly differ between groups.
Sharpe et al., 1997 (Cowen group)	10M CFS + neurasthenia 10M healthy	D-fenfluramine (30 mg, orally)	CFS: 1.6 Healthy: 1.1
Ottenweller et al., 2001 (Natelson group)	17F CFS (Holmes and Fukuda) 14F healthy	Maximal exercise test on treadmill	CFS: 1.2 Healthy: 2.2	Females tested in luteal phase. ACTH, EPI, TSH, DHPG, DOPA, and DOPAC responses were lower, GH response was higher, and NE response not significantly different.
Vasallo et al., 2001 (Cowen group)	16F, 4M neurasthenia (all but one also fulfilled Oxford and CDC criteria) 15F, 5M healthy	m-CPP (0.25 mg/kg, orally)	Only average difference from baseline reported, but not average prolactin values, thus increase ratio could not be calculated. However, there was not a significant difference between groups.	Females tested during early follicular phase. Participants also underwent a placebo test which did not raise prolactin in either group.
Weaver 2010 (Natelson group)	15F, 7M CFS (Fukuda) 8F, 3M CFS (Fukuda) + FM 10F, 6M healthy	L-tryptophan (120 mg/kg lean body mass)	CFS: 2.3 (females) and 2.2 (males) CFS+FM: 2.0 (females) and 3.0 (males) Healthy: 1.9 (females) and 2.2 (males)	Females tested during late follicular phase (days 7–14).

F: Females; M: Males; PVFS: Postviral fatigue syndrome; CFS: Chronic fatigue syndrome; ME: Myalgic encephalomyelitis; ACTH: Adrenocorticotropic hormone; TSH: Thyroid-stimulating hormone; GH: Growth hormone; NE: Norepinephrine; EPI: Epinephrine; EBV: Epstein-Barr Virus; Fukuda: Fukuda et al., 1994; Oxford: Sharpe et al., 1991; Holmes: Holmes et al., 1988; FM: Fibromyalgia

*For consistency between studies, prolactin response was calculated from group averages as peak prolactin divided by prolactin at timepoint 0. Data were approximated from plots using WebPlot Digitizer 4 when exact values were not reported.

Bolding indicates that the group was significantly different from the healthy control group for prolactin response. Note that the specific definition of response for the purpose of significance testing varied by study, and was not necessarily based on the definition of prolactin response as shown in this table.

Explanations for why I changed some parts:

Many of the prolactin response values changed, as I switched to use a consistent calculation method for every study (highest average prolactin value for each group divided by average baseline prolactin).

I removed the parts about timespan of testing and when specifically peak occurred. While it's interesting, I don't think it's that important for the purpose of the paper, so might make the table too busy.

Removed "Single author, a GP not an academic" for Richardson's paper because the single author factor can be determined from the intext citation which doesn't have et al, and the "not an academic" is I think not important enough to need to mention.

Removed "Results never published" since I think that's implied by (thesis) being shown next to the author name.

I didn't include the second Richardson paper at all (Richardson & Costa, 1998) because the criteria make it sound like they recruited participants specifically based on them having a large prolactin response. It's not entirely clear, but if so, that would lead to biased, misleading values in the table.

For the tags like (Cowen group), (Wessely group), etc, there is a complication that makes it not perfectly straightforward to use. O'Keane was senior author on both (Bakheit et al., 1992) and (Cleare et al., 1995). Wessely was an author on both (Cleare et al., 1995) and (Bearn et al., 1995). But there are no authors in common between (Bakheit et al., 1992) and (Bearn et al., 1992). So it's hard to say how to assign a meaningful "group" to these three studies.

 

[forestglip]

I've been extracting the data from the graphs in the buspirone studies, it looks something like this. For Bakheit 1992 I took males only, Sharpe 1996 is also males only, the other studies have both sexes combined.

EDIT: there might be an error in the Sharpe 1996 original data plot, see the discussion in the comments below.

View attachment 33897

I was looking at this collection of plots and was wondering if we could figure out why the first two studies seem to show much bigger differences between groups than the others. I think it's possible this relates to menstrual phase and dosage. Here are the buspirone studies which include relevant details:

Study	Sex	Rough assessment of group difference	Menstrual phase	Dosage	Notes
Bakheit et al., 1992	Males	Much larger increase		60 mg
Bakheit et al., 1992	Females	Much larger increase	Luteal	60 mg
Richardson, 1995	83% females	Much larger increase	Luteal	50 mg
Sharpe et al., 1996	Males	Somewhat larger increase		0.5 mg/kg up to 45 mg Based on mean weight of ~80 kg in the study, the mean dosage would be 40 mg.
Behan et al., 1996	100% males in patient group. 50% males in control group.	Somewhat larger increase		60 mg	The control group with a larger proportion of females would be expected to have a larger prolactin response, leading to a smaller difference between groups.
Majeed, 1996	50% females	Somewhat larger increase	Follicular	60 mg

It looks like it might be important to test during the luteal phase, and also to have a high enough buspirone dosage. That's kind of interesting because Dinan et al., 1990, in a study of healthy females, found that the largest prolactin response to buspirone occurred during the luteal phase.

 

[ME/CFS Science Blog]

I used your table as a basis to make one that I hope to include in the review:

Excellent work!

The group tags probably aren't that important. The main thing to note is that several studies were students from working under Behan as your review already does.