Predictors of dopamine agonist resistance in prolactinoma patients

Background Surgical resection of prolactinomas resistant to dopamine agonists is frequently incomplete due to fibrotic changes of the tumour under pharmacological therapy. In order to identify a subgroup of patients who may benefit from early surgery, we thought to investigate possible predictive factors of pharmacological resistance of prolactinomas to dopamine agonists. Methods We retrospectively analyzed a database of a Belgian tertiary reference center for patients with pituitary tumours from 2014 to 2016. The groups of interest were patients with dopamine agonist responsive and resistant prolactinomas. The possible predictive factors, including MRI findings, endocrinological parameters, response of tumour and patient factors for dopamine agonist resistance were investigated. Results We included 69 patients of whom 52 were women (75,4%) and 17 were men (24,6%). Rate of dopamine agonist resistance was 15.9%. We identified four significant predictors of dopamine agonist resistance: male gender, a large tumour volume, prolonged time to prolactin normalization and presence of a cystic, hemorrhagic and/or necrotic component. In addition, symptoms due to mass effect, high baseline prolactin level and a high contrast capture on MRI are factors that can be taken into consideration. Conclusion We identified predictive factors for pharmacological resistance and developed a scoring system for patient specific prediction of resistance to dopamine agonists. This scoring system may have impact on the timing and decision of surgery in prolactinoma patients after further prospective evaluation.

necrotic component. In addition, symptoms due to mass effect, high baseline prolactin level and a high contrast 23 capture on MRI are factors that can be taken into consideration. 24 Conclusion: We identified predictive factors for pharmacological resistance and developed a scoring system for 25 patient specific prediction of resistance to dopamine agonists. This scoring system may have impact on the timing 26 and decision of surgery in prolactinoma patients after further prospective evaluation.  49 tumour volume, a reasonable definition is regarded as 50 the failure to achieve prolactin normalization and/or a 51 tumour size reduction in coronal surface of ≥50%. The 52 definition only applies after a minimum period of 3 53 months of receiving a daily dose of 15 mg bromocriptine 54 or a weekly dose of 3.0 mg of cabergoline if tolerated [2]. 55 Dopamine agonist resistance occurs in patients with mi-56 cro-and macroprolactinoma, in 5 and 20% respectively 57 [2, 3, 6, 7]. Despite the fact that the side effects of DA 58 are limited, the intolerance rate is estimated to be ap-59 proximately 3-12% [2,8]. Drug resistance and intoler-60 ance, together with patient's preference, diagnostic 61 uncertainty and complications such as tumour apoplexy, 62 visual impairment and cerebrospinal fluid (CSF) leakage 63 due to shrinkage of the tumour are indications for trans-64 sphenoidal surgery [8]. The remission rate of surgery is 65 approximately 73-90% in microprolactinoma cases and 66 33-56% in macroprolactinoma cases [3,8]. The side ef-67 fects of surgery are rare and can be divided in two 68 groups: the minor (3.5-6.5%, e.g. septal perforation, epi-69 staxis, wound infection, hematoma, CSF leak and dia-70 betes insipidus) and major (1.5%, e.g. vascular injury/ 71 stroke, meningitis/abscess, visual loss and oculomotor 72 injury) complications [2,8]. 73 When resistance (failure to achieve PRL normalization 74 and/or a tumour size reduction of ≥50%) is established 75 to a particular patient, there are different therapeutic op-76 tions [9]. One of the options is to switch to another DA 77 since there is clear evidence that the switch to cabergo-78 line can overcome resistance to bromocriptine, with a 79 normalization of PRL and tumour mass reduction in 80 80 and 70% of the cases respectively [3, 6, 8]. Another ap-81 proach is a step-up dose augmentation if tolerated (side-82 effects are too prominent) and given a continued re-83 sponse. Although partial resistance may be suppressed 84 by a gradual increase of the dosage of dopamine ago-85 nists, it seems that for cabergoline (regarded as the most 86 efficient dopamine agonist) there is little benefit when 87 increasing the dose above 3.0 mg per week if continued 88 for at least 3-6 months [6, 10]. Another option is trans-89 sphenoidal surgery, which is widely considered as the 90 next gesture in cases of DA resistance.

91
However, second line surgical resection of dopamine 92 agonist resistant prolactinomas is frequently incomplete 93 because of fibrotic changes of the tumour due to dopa-94 mine agonists [11,12]. Fibrosis and uneven shrinkage of 95 the tumour can establish after 6 weeks of bromocriptine 96 treatment [11,12]. A recent study showed that 77% of 97 the prolactinomas with bromocriptine pretreatment 98 were fibrotic. The probability of fibrosis (22%) after 1 99 month cabergoline treatment was lower but still present 100 [13]. Finding fibrosis at the time of surgery is considered 101 as a negative predictive factor for complete biochemical 102 remission after surgery (0% versus 37%) [1,8,13,14]. As  The biochemical data we determined, were the levels 160 of PRL, insulin like growth factor -I (IGF-I), thyroid 161 stimulating hormone (TSH) and adrenocorticotropic 162 hormone (ACTH) and we quoted the presence of sex 163 hormone deficiency defined as decreased gonadotropins 164 or testosterone. 165 Furthermore we examined biochemical data as well as 166 different MRI variables such as the volume of the 167 tumour (h x l x w x (π/6)); the shape of the tumour (being 168 a sphere or bifocal); the intensity factor which is calculated 169 as the ratio between the intensity of the tumour and the 170 intensity of the grey matter (displayed in pixel value, PV), 171 always consequently at the level of the superior temporal 172 gyrus. Contrast enhancement, which is the ratio of the 173 density factor on T1 after contrast to the value on T1 174 without contrast, was likewise computed. We also investi-175 gated parameters for follow up regarding the evolution of 176 the tumour volume and the prolactin level.

177
The dosage of cabergoline was monitored during treat-178 ment in order to evaluate the response.

179
The dopamine agonist dosing regimen was clinically 180 adjusted and examined as follows. Effect of DA treat-181 ment was re-evaluated at least every 3 months. If the 182 treatment goal was achieved in terms of prolactin level 183 and tumour size, that same dose would have been con-184 tinued. If treatment goal was not achieved, the dose 185 would be augmented after evaluation by the endocrinol-186 ogists at the consultation.

187
When surgery was involved, a histological examination 188 of the surgical specimen was always performed. Informa-189 tion about the Ki 67 (proliferation marker) level and the 190 presence of sclerosis was hereby obtained. Sclerosis 191 (augmentation of fibrous tissue) is considered negative 192 when there was no mention of fibrosis, connective tissue 193 or sclerosis in the operation report; and when it is not 194 defined by the anatomy pathologist in the histological 195 examination report of the surgical specimen.  However, in that case, if tolerated and with the patient 203 consent, the dosage would further be increased. The re-204 sponse was monitored throughout the follow-up (at least 205 12 months) and changed if there was a response after 206 that further dose escalation.

207
Thereafter; a comparative study between the 2 groups, 208 responsive versus resistant prolactinoma patients was 209 performed.   (Table   T2 2). Causes of resistance in this second group 276 were: absence of prolactin normalization (5/11), < 50% 277 tumour volume shrinkage (1/11) or the failure of both 278 hormonal and tumour response (5/11). The demography 279 of the 2 groups was mainly different in terms of gender 280 with more men found in the resistant group compared 281 to the responsive group (45.5% vs. 20.7% respectively; 282 p = 0.08).

283
Although, the presence of symptoms was evenly dis-284 tributed among both groups, the isolated occurrence of 285 visual defects was only seen in case of resistance.   Mood's median test, 95% CI) f1:3 326 These are scored using a specialized scoring system 327 (Fig.   F2 2). Weaker predictors such as the presence of visual 328 defects, a high baseline prolactin level and a high con-329 trast enhancement on MRI, can also be taken into ac-330 count for the clinical decision. In our study population 331 itself, 89.5% could be correctly classified on the basis of 332 the scoring system if the cut-off of 20 points was 333 considered.
334 Discussion 335 In the present retrospective study, we thought to investi-336 gate possible predictive factors of pharmacological resist-337 ance of prolactinomas to dopamine agonists.

338
As expected, we found a higher general frequency of 339 microprolactinomas compared to macroprolactinomas. 340 There was a higher incidence of macroprolactinomas in 341 men (88.2%) compared to women (30.8%). This trend is 342 by analogy with previous literature [16][17][18].

343
The correlation we found between the (isolated) pres-344 ence of visual defects and resistance was confirmed in 345 literature, although they did not investigate the associ-346 ation with other combined mass effects such as head-347 aches [15]. We have only found a significant difference 348 between both groups when comparing patients who 349 merely had visual defects at time of diagnosis.  The proposition that resistance to dopamine agonists 366 is seen more frequently in men has previously been ac-367 knowledged in literature and is commonly adopted in 368 clinical practice [16,17]. Some authors believe that this 369 is due to a higher incidence of macroadenomas in men.  The strongest endocrinological predictor of resistance to 383 dopamine agonists in our study appeared to be the time to 384 prolactin normalization. Since there is a clearly defined dos-

392
In contrast to what could have been expected from lit-393 erature, we found no significant difference in baseline 394 prolactin level between the two groups, nor did we find