We conducted a 6-year retrospective study of patients with macroprolactinomas who were treated with CAB as a primary drug at Severance Hospital, Seoul, Korea between 2008 and 2013. Macroprolactinoma was defined as (1) a PRL level of at least 150 ng/mL and (2) a maximal diameter of at least 1 cm on baseline magnetic resonance imaging (MRI) scans of the sellar area. To evaluate the relationship between early and late parameters, the following additional inclusion criteria were applied: (1) a full dataset of pituitary hormone assays (including PRL) and sella MRI at baseline, (2) follow-up PRL assay and MRI after 3 months of CAB treatment, and (3) total follow-up duration of at least 12 months.

In all cases, oral CAB was started at a low dose (1 to 1.5 mg/wk), and the dosage was gradually increased to 1.5 to 4 mg/wk within 2 to 4 weeks. The increased dose was maintained for 3 months after treatment initiation to ensure a sufficient period of exposure to CAB. Evaluations of response occurred after 3 months, and the dose of CAB and interval of follow-up were continuously adjusted in consideration of response, tolerance, and other clinical indicators. For patients with good response and tolerance, reduction of CAB dose was carefully attempted with a relatively long interval of follow-up. For patients with poor response, a higher dose of CAB and a short interval of follow-up were employed. Because the follow-up intervals and treatment durations of the patients varied, three representative time points of response assessment were retrospectively defined as follows: early assessment, when the first sella MRI and PRL assay were performed after 3 months of CAB treatment; late assessment, when the first sella MRI and PRL assay were performed 12 to 24 months after treatment; and last assessment, when the last sella MRI and PRL assay were performed. Tumor volume was calculated according to Di Chiro and Nelson's formula: volume=height×length×width×π/6 [17]. The degree of response was assessed using TVR and NP. When evaluating the relationship between early response and later outcomes, we used group criteria according to the early response to CAB, as suggested in our previous article [16]: group 1, TVR ≥25% with NP; group 2, TVR ≥25% without NP; group 3, TVR <25% with NP; and group 4, TVR <25% without NP [16]. A successful response at the late or last assessment was defined as follows: volume response, TVR ≥50% without surgery; PRL response, NP without surgery; and CR, volume response with PRL response.

Serum PRL levels were measured via chemiluminescence immunoassay using commercial kits (Beckman Coulter, Brea, CA, USA). The within-run and total coefficients of variation for PRL concentrations were 3.66% and 3.77%, respectively. PRL levels <15 ng/mL for males and <25 ng/mL for females were regarded as normal. If serum PRL levels were normal or mildly elevated, PRL was measured again in diluted serum samples to exclude the hook effect, which can cause falsely low results [18,19].

Data were presented as the median (interquartile range) or the mean±SD. The relationship between early response and late or last response was analyzed using the Pearson correlation coefficient and multiple linear regression tests. The Student t test, Kruskal-Wallis test, Mann-Whitney U test, and Fisher exac t test were used to compare multiple groups. The relationship between tumor volume or PRL levels at baseline and TVR, PRL levels, or the maintained dose of CAB after treatment was analyzed using the Spearman correlation coefficient. Statistical analyses were performed using SPSS version 20 (IBM Co., Armonk, NY, USA). Values of P<0.05 were considered statistically significant.

Of 66 patients with macroprolactinomas who were treated with CAB as a primary drug during the study period, 47 patients were followed for at least 1 year from the initial administration of CAB to the end of the data collection. One patient was lost to follow-up before 12 months, and two female patients who became pregnant less than 12 months into CAB treatment were also excluded because CAB had to be discontinued irrespective of tumor size and PRL level. Finally, in total, 44 patients, including 28 males (63.6%), were included in the study (Tables 1, 2). The mean age of the patients was 36.8 years, and the median follow-up duration was 16 months (interquartile range, 15 to 25.5). Eleven patients (25%) had visual field defects, and 28 patients (61.4%) complained of sexual dysfunction including impotence, oligomenorrhea/amenorrhea, and/or infertility. The patients also complained of headache and dizziness (n=12, 27.3%), ocular movement abnormalities (n=2, 4.5%), and galactorrhea (n=6, 13.6%). The median PRL level and median tumor volume were 796.7 ng/dL (range, 202.5 to 2,431.3) and 3.71 cm³ (range, 1.60 to 11.51), respectively. Tumor invasion of the cavernous sinus was noted in 20 patients (45.5%). Most patients (n=40, 90.9%) displayed sex hormone deficiency, defined as a testosterone level below the lower limit of the normal population for males, oligomenorrhea/amenorrhea in premenopausal females, and an inappropriately low gonadotropin level in postmenopausal females. Five patients (11.4%) displayed growth hormone (GH) deficiency, defined as a lower serum insulin-like growth factor 1 (IGF-1) level than the age- and sex-specific lower limit of the normal population, whereas secondary hypothyroidism and adrenal insufficiency were found in one patient and zero patients, respectively. Compared with female patients, male patients were older (41.5 years vs. 28.5 years, P<0.001), less likely to complain of sexual dysfunction (46.4% vs. 87.5%, P=0.010), and more likely to have a visual field defect (39.3% vs. 0%, P=0.003), and they displayed higher tumor volume (10.64 cm³ vs. 0.87 cm³, P<0.001) and PRL level (923.3 ng/mL vs.428.5 ng/mL, P=0.023).

At the last assessment, a volume response, PRL response, and CR were achieved by 35 (79.5%), 36 (81.8%), and 31 patients (70.5%), respectively (Table 1). Two patients who exhibited transient mild PRL elevation (patients no. 18 and 24, who had PRL levels of 52.3 and 33.6 ng/mL, respectively) showed very poor drug compliance at the last assessment, but were regarded as having achieved a PRL response. Four patients (9.1%) underwent surgery in their treatment courses, and the median TVR of the patients who did not undergo surgery was 74.7% (range, 61.7 to 85.4).

Neither the baseline tumor volume nor PRL level were directly correlated with TVR or PRL level at the early, late, and last assessments (Spearman correlation coefficients, all P>0.05). However, six patients with large tumors (larger group), whose baseline tumor volume exceeded 18 cm³ and whose PRL level exceeded 2,000 ng/mL, displayed higher PRL levels than patients with smaller tumors (smaller group) 3 months after treatment (22.9 ng/mL vs. 1.6 ng/mL, P=0.043) (Fig. 1A). Interestingly, these differences in PRL levels between groups lost their statistical significance at 9 months (Fig. 1A), and the TVR of patients who did and did not undergo surgery did not vary between the groups during the entire treatment course (Fig. 1B).

To assess the influence of early response on late or last response, we primarily categorized patients into four groups according to their early responses to CAB. Of the total 44 patients, 29 patients were included in group 1 (TVR ≥25% with NP), eight in group 2 (TVR ≥25% without NP), five in group 3 (TVR <25% with NP), and two in group 4 (TVR <25% without NP) (Fig. 2A). Most patients in group 1 (93.1%) achieved a CR as expected, but two patients (6.9%) did not achieve TVR ≥50% until the last assessment. Four of eight patients (50%) in group 2 eventually achieved a CR, whereas no patients in groups 3 and 4 achieved a CR. The proportion of CRs significantly differed between the groups (group 1 vs. 2 vs. 3 vs. 4; 93.1% vs. 50% vs. 0% vs. 0%; P<0.001). The last maintenance dose of CAB differed according to the last responses (CR vs. volume response without NP vs. PRL response without TVR ≥50% vs. surgery; 1.5 mg/wk (range, 1.0 to 2.0) vs. 4.5 mg/wk (range, 2.75 to 4.5) vs. 2.0 mg/wk (range, 1.0 to 2.0) vs. 0.5 mg/wk (range, 0.0 to 0.7); overall P=0.006), but no differences were observed between the four groups according to early responses (P=0.109).

We conducted further analysis of change in PRL levels to identify the distinguishing characteristics of the patients in group 2 that eventually achieved a CR (Fig. 2B). Of the four patients who achieved NP, three patients displayed normal PRL levels at 6 to 9 months (patients no. 32, 33, and 37). The remaining patient, who had a giant prolactinoma (patient no. 31; maximum tumor diameter 4.8 cm, tumor volume 20.4 cm³, and PRL level 8,759 ng/mL at baseline), maintained a slightly supranormal range PRL until 15 months (18.6 ng/mL at 9 months and 19.5 ng/mL at 15 months), but exhibited NP at 21 months. Median PRL levels did not differ between the patients who achieved NP (delayed response subgroup, n=4) and those who did not achieve NP (sustained resistance subgroup, n=4) until 3 months (1,947.4 ng/mL vs. 813.5 ng/mL, P=0.886 at baseline; 55.3 ng/mL vs. 115.3 ng/mL, P=0.686). However, a meaningful difference in median PRL levels between the two subgroups could also be observed after 9 months of CAB treatment (median PRL level: 13.0 ng/mL vs. 117.9 ng/mL, P=0.029). Median TVR did not vary between the subgroups (59.1% vs. 65.3%, P=0.486 at 3 months; 81.9% vs. 74.7%, P=0.200 at 15 months).

We also attempted to determine whether the absolute value of early TVR itself is predictive of response or whether only a certain cutoff such as 25% is meaningful. Pearson correlation analysis demonstrated that late TVR had a very strong correlation with early TVR (R=0.869, P<0.001) according to the following regression equation: late TVR=0.851×early TVR+26.606 (Fig. 3). Last TVR also displayed a correlation with early TVR, although the strength of the correlation was slightly lower (R=0.785, P<0.001). Multiple regression analysis was subsequently performed using sex, natural logarithm of the CAB maintenance duration between early and late assessment (Ln [weeks]), and cumulative dose of CAB between early and late assessments (mg) as independent variables. In this analysis, a greater TVR at the early assessment and a longer CAB maintenance duration were independent predictors for greater TVR at the late assessment (β=0.849, P<0.001; and β for Ln [weeks], 16.978, P=0.016, respectively), whereas sex and cumulative dose of CAB were not predictive (P=0.074 and P=0.613, respectively).

Because the patients who underwent surgery maintained the lowest dose of CAB (Fig. 2A), we sought to confirm whether surgery could independently reduce CAB dose. Thus, we performed multiple linear regression analysis with four independent variables as follows: undergoing surgery, baseline PRL level (mg/mL), 1/baseline volume (1/cm³), and natural logarithm of the treatment duration (Ln [months]). Baseline volume was transformed reciprocally to ensure the linearity of the model. Analysis confirmed that undergoing surgery (β=-1.181, P=0.013), a lower baseline PRL level (β=0.161, P=0.008), a smaller baseline volume (β for 1/baseline volume=-0.448, P=0.026), and a longer treatment duration (β for Ln [months]=-0.882, P=0.010) were independent predictors for a lower dose of CAB at the last assessment.

Four patients underwent transsphenoidal surgery (Table 3; patient nos. 38, 42, 43, and 44). The first patient (patient no. 38) underwent surgery after 3 months of CAB treatment because of acute hemorrhage in the tumor accompanied by severe headache. Immunohistochemical analysis for all pituitary hormones including PRL was negative, but the PRL result was regarded to be false-negative due to destruction of tumor cells following acute hemorrhage, as a nonfunctioning pituitary adenoma could not explain the patient's extremely high baseline PRL level (2,042 ng/mL). His headache resolved completely after surgery, although he developed panhypopituitarism. His CAB dose was reduced from 3 to 0.7 mg/wk, and his serum PRL levels were within the reference range. The second patient (patient no. 43) underwent surgery at 5 months because of a progressive visual field defect and diplopia. Immunohistochemical analysis of the excised tumor revealed PRL immunoreactivity. His symptoms were dramatically improved after surgery, although panhypopituitarism developed. The patient was able to discontinue CAB with sustained NP, and MRI revealed no residual tumor. The third patient (patient no. 42) underwent elective surgery after 12 months of CAB treatment owing to CAB resistance. Her early TVR was only 5.7%, and her preoperative PRL level was 168.6 ng/mL. Immunohistochemical analysis demonstrated PRL reactivity. She was able to discontinue CAB with sustained NP and no evidence of a residual tumor. The fourth patient (patient no. 44) also underwent elective surgery after 16 months of CAB treatment due to CAB resistance. His tumor size had increased by 2.1% despite 3 months of CAB treatment, and it continued to grow after 12 additional months of treatment without NP. Immunohistochemical staining revealed immunoreactivity for PRL. Postoperative MRI could not confirm whether there was residual tumor or only postoperative changes. PRL levels were not normalized despite surgery; however, CAB dose was markedly decreased (from 3 to 1 mg/wk) with a partial reduction in serum PRL level (from 699.4 to 393.8 ng/mL).

Overall, none of the patients achieved TVR ≥25% at 3 months, including those with (patient no. 38 and 42) and those without (patient no. 43 and 44) NP. CAB dose was decreased after surgery, and two patients were able to discontinue CAB without evidence of residual tumor or PRL elevation.

Among the patients who did not undergo surgery, CAB treatment improved baseline abnormalities of other pituitary hormones in most. All subnormal IGF-1 levels (patient no. 6, 8, 9, 13, and 24) and secondary hypothyroidism (patient no. 34) were gradually restored. All premenopausal women with oligomenorrhea/amenorrhea demonstrated improvement of symptoms following reduction of PRL levels, and the subnormal gonadotropin level of one postmenopausal woman was also improved. However, six male patients with NP and two male patients without NP displayed subnormal serum testosterone levels at the last assessment.

To elucidate the early characteristics of patients who could not achieve normalization of testosterone levels despite NP, we compared various clinical parameters of these patients (subnormal group, n=6) with those of male patients with normal testosterone levels and NP at the last assessment (normal group, n=16). Three patients who underwent surgery and one patient who received exogenous testosterone replacement before the first visit to our institution were excluded from this analysis to minimize confounding factors. The subnormal group had lower baseline IGF-1 (median 112.0 ng/mL vs. 212.1 ng/mL, P=0.013) and testosterone levels (median 57.9 ng/dL vs. 159.4 ng/dL, P=0.010) than the normal group. Patients in the subnormal group tended to be older (48.4 years vs. 39.8 years, P=0.052), and this group had a higher proportion of patients with age-specific subnormal IGF-1 levels (50% vs. 6.3%, P=0.046). Baseline tumor volume, PRL levels, proportion of patients with symptoms related to volume effects, and responses to CAB in terms of TVR and residual volume did not differ between the groups.