Visual Loss in Pregnant Women with Pituitary Adenomas

Methods

Patients

We examined the records of all women with pituitary adenomas who had been evaluated by the New York University neuro-ophthalmology service between 1982 and 1992 and who became pregnant. Sixty-six patients became pregnant who had not had previous surgery or radiation therapy directed at the adenoma (mean age at pregnancy, 30.6 years; range, 21 to 38 years). Except for one patient (patient 3), a diagnosis of pituitary adenoma had been established before referral by endocrinologists or gynecologists for consultation and baseline visual function testing (see vision analysis) during the first trimester or before conception. During pregnancy, a neuro-ophthalmologist (MJK) prospectively repeated visual testing at least once, usually in the third trimester. Missing endocrine data and neuroimaging studies for review were sought by mail and telephone inquiries with the assistance and consent of the patients and referring physicians for all patients.

Neuroimaging Analysis

High-resolution computed tomography or magnetic resonance imaging (or both) were done in all but two patients before the first pregnancy. Coronal- and axial-oriented images (2.5- to 5.0-mm thick) for computed tomography studies and coronal-, axial-, and sagittal-oriented images (3-mm thick) for magnetic resonance imaging were all evaluated for the presence of an adenoma. On the coronal images, the maximum vertical height (the direction of growth toward the optic chiasm and typically the largest dimension) of the tumor mass was measured.

Eight patients had adenomas that were smaller than 0.3 cm and difficult to distinguish from the surrounding pituitary gland. These patients were considered to have an adenoma because of abnormal tilting of the pituitary stalk, asymmetric fullness of a pituitary hemisphere, amenorrhea, infertility, and high levels of serum prolactin. The tumor height in these patients was considered to be zero for the purpose of data analysis.

In general, patients with adenomas of 1.0 cm or larger and no visual loss were counseled to avoid pregnancy unless they had previous treatment to decrease the probability of optic structure compression. This may partially explain the small number (eight) of patients with adenomas larger than 1.1 cm in this series.

Vision Analysis

We prospectively collected data on the visual and ocular motor systems. Before pregnancy, 55 patients had a normal baseline neuro-ophthalmologic examination. Within 3 weeks of the calculated conception date, 10 additional patients were found to have normal vision. Only 1 patient was not examined until the end of the second trimester (patient 3). This patient was included because she had a normal ophthalmologic examination, done by the referring ophthalmologist, 1 year before her pregnancy. The evaluation included the best corrected visual acuity; color vision testing using pseudoisochromatic plates; the evaluation of lid, pupil, and ocular motility functions; and ophthalmoscopy. Visual field measurement was done with tangent or Goldmann perimetry in 52 patients and, in later years (after 1987), with computerized threshold perimetry in 14 patients.

Each patient was encouraged to return at yearly intervals if she was not pregnant and during the first trimester if she was pregnant. Patients were examined at least one other time, usually in the last trimester, or more often if a potential problem occurred (such as frequent headaches, visual loss, or concern by the patient or referring doctor). Patients were also instructed to be examined before having a subsequent pregnancy. Visual loss was diagnosed when visual field defects typical of optic nerve or chiasm compression occurred (Figure 1)—not when minor degrees of nonspecific threshold elevation were present—with or without a decrease in acuity or color vision.

Endocrine Studies

The patients had no other disorders and were receiving no drugs known to influence prolactin levels at the time of diagnosis [12]. Serum prolactin levels were measured in all 65 patients in the analysis and were increased in 58 (range, 40 µg/L to 2000 µg/L; normal, 0 to 25 µg/L). The prolactin level was greater than 100 µg/L in 27 patients. Two patients had acromegaly with serum growth hormone levels of 55.2 µg/L and 81.2 µg/L (normal <8 µg/L) and prolactin levels of 7.8 µg/L and 69.8 µg/L (patients 6 and 8, respectively). Five patients had data consistent with nonsecretory tumors.

In patients diagnosed as having a prolactinoma, bromocriptine mesylate (Parlodel; Sandoz Pharmaceuticals Corporation, East Hanover, New Jersey), 2.5 mg one to three times daily, was prescribed. Although some patients admitted to not taking the bromocriptine, an accurate rate of compliance was not determined. Bromocriptine was stopped at a mean of 4.45 weeks from the calculated date of conception in 56 pregnancies (range, 2 days to 23 weeks). Neither patient with acromegaly received medication before or during pregnancy.

The presence of a clinical pregnancy reported by each patient was confirmed by a conventional laboratory pregnancy test [13]. Failure to sustain the pregnancy to term was considered a spontaneous abortion. Examination for fetal tissue was not uniformly done.

Data Analysis

One patient with a 2.0-cm tumor conceived three times and immediately had medical abortions each time because of her fear of visual loss. Because all her pregnancies were terminated, we omitted data from this patient, leaving 65 patients in the final data analysis.

Data were analyzed using simple logistic regression analysis for yes-no outcomes and using simple regression analysis for continuous outcomes. The small sample size precluded the use of any multiple regression procedures. Linear regression [14] data were interpreted by the size and significance of their β-coefficients, whereas the logistic regression data [15] were assessed by examination of odds ratios. The odds ratio in such patients was obtained using the β-coefficients (an exponential calculation). Confidence intervals (95% CI) accompanied the odds ratio or β-coefficient for each condition. A two-tailed Fisher exact test was used to calculate exact P values. All statistical calculations were done using SAS [16].

Results

A total of 111 pregnancies was noted. The neuro-ophthalmologic examination showed visual field defects from tumor compression of the anterior visual pathway in six patients during pregnancy (Table 1); five of these patients were in their first pregnancy (patient 3 was first seen during the second trimester). The visual field defects occurred in one eye in one patient and in both eyes in five patients (Figure 1). An incomplete superior temporal quadrantanopia was found in eight eyes of six patients. A complete superior temporal quadrantanopia was found in two eyes of two patients. An incomplete temporal hemianopia (progressed from a partial quadrantanopia) was found in the one patient (patient 3) with decreased visual acuity in one eye. Concomitant with the field loss, the visual acuity was 20/30 or better in eyes from all patients except for one eye of one patient (patient 3), which was 20/80. When patient 3 was first examined by the neuro-ophthalmologist (MJK), the acuity of this eye was 20/30.

The color vision was normal in all the eyes except in the 1 patient with decreased acuity in one eye. All of the patients having visual loss had adenomas of 1.2 cm or more. No women developed dysfunction of cranial nerves III, IV, V, or VI, clinical signs of an infarction, or hemorrhage in the pituitary gland or adenoma. No patients had any other signs of central nervous system dysfunction or increased intracranial pressure.

Treatment was individualized for each patient with visual loss (Table 1). When treated postpartum with bromocriptine, two patients had normal visual examinations within 3 months. Bromocriptine administered during the last 5 months of pregnancy normalized the vision in one patient (within 3 months). Bromocriptine therapy combined with a therapeutic abortion in one patient led to a return of normal vision (within 1 month). Trans-sphenoidal adenomectomy, done in the third trimester in patient 3 and done postpartum in a second patient, resulted in normal visual function (visual acuity > 20/30, normal color vision, full visual fields) within 4 months.

Eight patients had adenomas with a vertical height of 1.2 cm or more. Six of these women with large tumors composed the group who developed the described visual loss (Table 1). Because 100% of the patients with visual loss had large adenomas, the odds ratio for developing visual loss during pregnancy in relation to increasing tumor size (per cm of vertical height) on the baseline scan could not be determined. None of the 57 patients with adenomas of maximum vertical height of 1.0 cm or less (mean, 0.43 cm; SD, 0.35) had visual loss. When data from the eight patients with adenomas smaller than 0.3 cm were eliminated, the mean vertical tumor height was 0.65 cm (SD, 0.26).

Ten patients, 9 with adenomas smaller than 0.5 cm and 1 with a 1.0-cm adenoma, had three or more pregnancies. None had visual loss during any pregnancy. In this group, 9 patients had prolactinomas (baseline prolactin range, 34 to 305 µg/L) and 1 had a nonsecretory adenoma. Bromocriptine was used to induce 14 pregnancies in this group of women. The development of visual loss could not be correlated with the baseline serum level of prolactin or the occurrence of multiple pregnancies.

Therapeutic abortions were done during nine pregnancies. A tubal or ectopic pregnancy occurred in two pregnancies that were included among the spontaneous abortions. Spontaneous abortions (15 of 55 pregnancies) occurred in 12 of 33 (36%) patients who had not taken bromocriptine before the pregnancies (4 in one patient). Spontaneous abortions (4 of 56 pregnancies) occurred in 4 of 44 (9%) patients treated with bromocriptine (difference, 27%; 95% CI, 9% to 46%; P = 0.005).

Discussion

Among patients with a pituitary adenoma of vertical height of 1.2 cm or more, six developed visual loss during pregnancy. However, because we eliminated or did not consider those patients who had previous surgery, the estimate of risk is uncertain. We are more confident of our estimate of risk in patients with microadenomas or tumors smaller than 1.1 cm. During 103 pregnancies, none of the 57 patients (CI, 0% to 6.3%) in this group developed visual loss.

Although we cannot exclude the idea that occurrence of visual loss might be as high as 6.3%, our results appear to differ from those of previous studies [2, 17] that described a 5% incidence of visual loss occurring during pregnancy in patients with a “microadenoma.” Our finding is not truly comparable to previous conclusions because earlier results were based on data from skull radiographs, tomography, or early computed tomographic scans. Additionally, in previous studies, visual status was not consistently examined. Inferences drawn from these imaging techniques such as changes in the shape and intactness of the bony sella cannot accurately assess the size of a pituitary adenoma. In addition, macroadenomas that grew upward without widening the sella or destroying the floor may have been misclassified as microadenomas.

Imaging and measurement of adenoma size improved with the high-resolution, thin-slice computed tomography and coronal orientation imaging used in this study. However, adenoma dimensions can be more accurately determined with high-field magnetic resonance imaging using multiple planes and intravenous gadolinium. Thus, all patients suspected of harboring a pituitary tumor, particularly if a pregnancy is anticipated, should have high-resolution, thin-slice imaging of the pituitary region with either computed tomography or magnetic resonance imaging.

Harboring an untreated pituitary adenoma appears to increase the risk for miscarriage, with an incidence of 27% in our study and an incidence as high as 32% in other reports [1, 7]. A decrease in the miscarriage rate was noted for patients who conceived while they were receiving bromocriptine, even though the drug was stopped early in the pregnancy. The incidence (7%) of spontaneous abortions in our patients treated with bromocriptine was comparable to the incidence (13%) of miscarriages for clinically recognized pregnancies in normal women [13] and the incidence (11.1% to 17%) previously reported [2, 9, 18] in women with hyperprolactinemia who were treated with bromocriptine. Multiple pregnancies might predispose patients to visual loss because of recurrent exposure to factors that can increase the size of the normal pituitary gland, but our data do not support this speculation [19]. Despite having three or more pregnancies, no patient with an adenoma of 1.0 cm or smaller had visual dysfunction.

Depending on whether visual loss develops during pregnancy, several therapeutic options exist for patients with prolactin-secreting macroadenomas. Patients without visual loss can have frequent monitoring of their visual status (every 3 months) or can be treated with a dopamine agonist during pregnancy as a preventive measure [20, 21]. Patients in whom visual loss develops will likely require bromocriptine during the remainder of the pregnancy [8, 22-24]. Others have favored prepregnancy prophylactic surgical removal of a macroadenoma, citing a 15% to 35% incidence of “serious tumor enlargement” during pregnancy when only bromocriptine is used [25, 26]. However, the failure of medical therapy in pregnant women has not been adequately shown.

A patient with an adenoma that is not prolactin secreting or is resistant to dopamine agonists because of a reduced number of dopamine receptors [27] requires a different approach. When such a patient has mild visual field loss, as occurred in one of our patients, the patient can be followed with frequent monitoring of visual function without direct intervention. However, if the visual loss is severe or progressive, trans-sphenoidal surgical decompression of the intracranial optic nerves and chiasm may be necessary during pregnancy. Another approach suggested by some investigators is to give radiation therapy before pregnancy as an alternative means of preventing visual loss [26, 28].

We found that patients with pituitary adenomas (≥ 1.2 cm) were at greater risk for developing visual loss during pregnancy than those patients who had microadenomas. All patients suspected of having a pituitary tumor, particularly if a pregnancy is anticipated, should have high-resolution, thin-slice imaging of the pituitary region with either computed tomography or magnetic resonance imaging.