Evaluation of percentage recovery together with modified reference range in hyperprolactinemia

Introduction

Human serum prolactin (PRL) is known to exist in three major molecular variants, which are monomeric PRL (little or free PRL, 23 kDa), dimeric PRL (big PRL, 50–60 kDa) and big-big PRL (macroprolactin, 150–170 kDa). Monomeric PRL is the main circulating form of PRL (approximately 80–95% of total PRL) and accounts for the majority of PRL bioactivity and immunoreactivity in the serum of most patients with hyperprolactinemia as well as healthy individuals. In normal sera, dimeric PRL and macroprolactin account for 10–15% and <1% of total prolactin, respectively [1]. Macroprolactin is considered to have minimal bioactivity, however, it retains immunoreactivity with varying degrees depending on the specificity of the assay used to measure PRL levels [2], [3].

Macroprolactinemia, increased levels of macroprolactin, is one of the causes of hyperprolactinemia which usually requires no treatment. However, there are no reliable clinical features or laboratory tools to differentiate patients with macroprolactinemia from patients with elevated levels of biological active monomeric PRL (true hyperprolactinemia) [4]. This may lead to concerns for both the patient and clinician, unnecessary drug treatments and interventions [5]. Therefore, when serum prolactin is high, it is recommended that laboratories screen for macroprolactinemia to discriminate between those patients with true hyperprolactinemia and those with macroprolactinemia [6].

The reference method for the measurement of high molecular mass forms of prolactin is gel filtration chromatography (GFC). But this method is impractical for routine performance in clinical laboratories because it is labor-intensive, time-consuming and expensive [7]. In recent years, several laboratories have been performing polyethylene glycol (PEG) precipitation method to determine the ratio of post-PEG PRL (PRL concentration after PEG treatment) to total PRL levels which is both cheap and easy to perform in clinical laboratories. The PEG precipitation technique is based on the removal of the macromolecular form of PRL by precipitation with PEG [7], [8], [9]. In this technique, PRL levels are measured both in the serum (total PRL) and in the supernatant, which is obtained after precipitation with PEG (post-PEG PRL). The ratio of post-PEG PRL levels to serum total PRL is identified as the percentage of recovery (R%). There are several studies, comparing the results of PEG precipitation with GFC [9], [10], that yielded promising results and R% <40–50% has been suggested as the threshold for the determination of macroprolactinemia [5], [11], [12], [13]. However, increased monomeric PRL levels may be present accompanied by macroprolactinemia. It is therefore critical not only to detect the presence of macroprolactinemia by using R% value but also to measure the amount of the monomeric PRL, as well. For this purpose, it is suggested to evaluate if the post-PEG PRL level fall in the range obtained from the PEG-treated supernatant of healthy individuals (modified reference range) to confirm the absence of true hyperprolactinemia accompanying with macroprolactinemia [5], [14], [15], [16].

This retrospective study aimed to examine the added value of the consideration of post-PEG PRL levels in the determination of macroprolactinemia and true hyperprolactinemia. For this purpose, we examined the macroprolactin reports of our laboratory in the last 35 months and created a modified reference range for the post-PEG PRL obtained from the data of healthy controls. We reevaluated data of patients with hyperprolactinemia considering this modified reference range.

Materials and methods

Results

The results obtained in hyperprolactinemic and normoprolactinemic subjects are summarized in Table 1. PRL recovery (R%) for the 390 patients with hyperprolactinemia assayed after PEG-precipitation ranged from 2 to 106%. The 95% reference ranges (2.5th–97.5th percentiles) for PRL in healthy subjects after PEG treatment ranged from 2.5–20.5 μg/L in females and 1.9–14.5 μg/L in males (Table 2).

Using the cut-off <40% for recovery, the prevalence of macroprolactinemia in hyperprolactinemic patients were 24.9%; whereas 67.4% of patients showed a true hyperprolactinemia (R>60%) and 7.7% were evaluated as suspicious (40%<R<60%). Figure 1 shows the recovery % frequency of distribution in hyperprolactinemic subjects. When we evaluated our data according to gender, macroprolactinemia was detected in 76 patients (24.7%) and true hyperprolactinemia was detected in 210 patients (68.2%) within 308 hyperprolactinemic female patients. 22 of these patients (7.1%) were evaluated as suspicious. However, in 82 male patients with hyperprolactinemia, 21 patients (25.6%) were identified as macroprolactinemia, 53 patients (64.6%) as true hyperprolactinemia and 8 patients (9.8%) were considered as suspicious. Accordingly, there was no statistically significant difference between genders in terms of macroprolactinemia prevalence (p=0.137).

Figure 1:

Recovery % frequency distribution in hyperprolactinemic subjects. According to the 40% recovery criteria, macroprolactinemic subjects, white columns; suspicious subjects, gray columns; true hyperprolactinemic subjects, black columns.

We also evaluated our reports according to the alternative cut-off (50%) for R% and we observed 116 patients (29.7%) were in the macroprolactinemia range. According to the modified reference range for post-PEG PRL, 105 of the 390 hyperprolactinemic cases (26.9%) were identified as macroprolactinemia, i.e. the post-PEG PRL levels fell to less than 20.5 μg/L in sera from women and to less than 14.5 μg/L in samples from men after treatment with PEG (Table 3).

On the other hand, the prevalence of macroprolactinemia in normoprolactinemic subjects was 2.3% (n=3), and 6.1% of participants (n=8) showed an R% value in the 40–60% range.

Comparisons were also made between two interpretive approaches (recovery criteria and modified reference range) for categorizing subjects as having macroprolactinemia or true hyperprolactinemia. Of the 390 cases with hyperprolactinemia, 84 cases were identified as having macroprolactinemia, and 257 cases were identified as having true hyperprolactinemia using either the modified reference range or percentage recovery approach (Figure 2). Thirteen of the 97 macroprolactinemic subjects (13.4%) according to the 40% criterion were considered to be hyperprolactinemic using the modified reference range. Total PRL levels of these subjects were >70 μg/L. Six of the 263 hyperprolactinemic patients (2.3%) who were reported as true hyperprolactinemia according to the 40% criterion were considered to be macroprolactinemic by using the modified reference range. Serum total PRL levels in these six patients were <30 μg/L. On the other hand, we found that 15 suspicious results (50%) ended with a normal post-PEG PRL concentration after PEG precipitation (macroprolactinemic subjects), whereas in the remaining 15 cases the values remained elevated (true hyperprolactinemic subjects). We investigated the consistency of 50% recovery criterion with the modified reference range. We observed that 21 patients (18.1%) identified as macroprolactinemia with the 50% criterion were in the range of true hyperprolactinemia and 11 true hyperprolactinemia cases (4%) with the 50% criterion were in the range of macroprolactinemia according to the modified reference range (Figure 3).

Figure 2:

Evaluation of the 40% recovery (R) results with the modified reference range (MRR)/MacroPRL, macroprolactinemia; hyperPRL, hyperprolactinemia.

Figure 3:

Evaluation of the 50% recovery (R) results with the modified reference range (MRR).

Discussion

This retrospective study demonstrated that, 24.9% of hyperprolactinemic samples reported by our laboratory in a 35 months period were identified as macroprolactinemia according to the 40% criterion. The ratio of macroprolactinemia in hyperprolactinemic subjects which is believed to be assay dependent, has been reported between 10 and 46% in several studies [4], [11], [12], [19], [20], [21]. All these data, including ours, indicate that macroprolactinemia is a common cause of hyperprolactinemia. In addition, we observed macroprolactinemia in 3 (2.3%) of the normoprolactinemic subjects which is comparable with some other reports [22], [23]. Hattori et al. [22], reported that the prevalence of macroprolactinemia in healthy population was 3.86% in women and 3.13% in men; those authors also reported that there were no differences between males and females. However, in our study, all of the three participants of normoprolactinemic subjects with macroprolactinemia were female but the number of samples is low to make any significant comments. We would like to point out that the significance of macroprolactinemia in normoprolactinemic subjects remains unknown.

Reference ranges for monomeric PRL levels in the post-PEG supernatant (post-PEG PRL) have been reported for widely used automated immunoassays [15], [16], [18]. We compared our modified reference ranges with the ones that were obtained by the Abbott Architect assay that we had been using (Table 2). Gender-specific reference ranges, particularly upper limits, were similar with those found by Whitehead et al. [18]. However, the upper limits of reference ranges were considerably higher than those established by Beltran et al. [15] and Overgaard and Pedersen [16].

The findings of this study revealed that the number of samples identified as macroprolactinemia would not significantly change (p=0.126), if the patients had been reported related to the modified reference range derived from normal subjects instead of 40% criterion (26.9% and 24.9%, respectively) as we had already been doing. However, it was observed that 13 subjects that had been identified as macroprolactinemia using 40% criterion had higher post-PEG PRL levels than those of the healthy controls. Since those patients should have been reported as true hyperprolactinemia accompanying with macroprolactinemia, there might be a latency in their treatment process. These subjects account for the 13.4% of macroprolactinemic patients reported according to the 40% criterion and had total PRL levels >70 μg/L. Consistent with this observation, McCudden et al. [24] suggested that when the PRL concentration was >85 μg/L clinicians should not be concerned with macroprolactinemia, but focus on routine evaluation of hyperprolactinemia. The coexistence of true hyperprolactinemia and macroprolactinemia emphasizes the importance of modified reference range for the post-PEG PRL [25]. On the other hand six subjects (2.3%) who were reported as having true hyperprolactinemia using 40% criterion were macroprolactinemic when evaluated with the modified reference range. Although R% classification of hyperprolactinemia has advantages and benefits in daily laboratory practice, it seems like that it is unsatisfactory to distinguish macroprolactinemia and true hyperprolactinemia when reports solely include R% criterion. Furthermore, there is a grey zone for R% (40–60%) in which the method has almost no efficiency and it is stated that those samples should be subjected to GFC for a definitive diagnosis [8]. In order to cope with this problem some authors also suggest 50% cut-off for R% [26]. When we evaluated our reports according to the cut-off 50%, we observed 116 patients were in the macroprolactinemia range however, according to the modified reference range 21 of these patients (18.1%) had true hyperprolactinemia. So, the ratio increased from 13.4 to 18.1% for the misdiagnosis of coexistence of true hyperprolactinemia and macroprolactinemia when we used 50% instead of 40% cut-off for recovery.

In conclusion, these data show that macroprolactinemia is a common cause of hyperprolactinemia, however discrimination capacity of R% criterion for true hyperprolactinemia and macroprolactinemia is limited, particularly when macroprolactinemia coexist with true hyperprolactinemia. We suggest that, in accordance with R% criterion, laboratory reports should include the post-PEG PRL levels along with the modified reference range of this component derived from healthy individuals. It is important that adding this data to their reports will not cause any expense and will not bring any extra workload since these values have already been obtained if the laboratory reports the R% criterion of macroprolactinemia.