Comparison of perimetric 24‐2 and 30‐2 test patterns in detecting visual field defects in patients with tumours in the pituitary region

Automated perimetry provides a standardized method of measuring the visual field. The Humphrey Field Analyser (HFA) uses the 24‐2 test pattern to cover 24 degrees centrally or the 30‐2 test pattern to cover a slightly broader region of 30 degrees. The aim of this study was to determine whether the 24‐2 test pattern provides comparable information to the 30‐2 test pattern in detecting visual field defects in patients with tumours in the pituitary region.


| I N T RODUC T ION
The visual pathway in the pituitary region is a vulnerable part of the visual sensory system.The chiasm and adjacent parts of the optic nerve and the optic tract can be affected by several types of lesions, most commonly by pituitary tumours, causing various types of visual field defects (Abouaf et al., 2015;Boland et al., 2016;Danesh-Meyer et al., 2019).Pituitary adenomas are the second most frequent benign intracranial tumours, and account for approximately 10-15% of all intracranial neoplasms (Chen et al., 2021;Miller et al., 2021).
Surgery is indicated when a lesion exerts a mechanical compressive effect on sensitive surrounding structures, for example the visual pathways or cranial nerves within the cavernous sinus and brain parenchyma, leading to visual loss, endocrine dysfunction or other neurological symptoms (Theodros et al., 2015).Visual dysfunction is reported to be the main presenting symptom in patients undergoing surgical decompression and is reported in as many as 38-72% of the patients (Carrim et al., 2007;Ebersold et al., 1986;Ogra et al., 2014).Most patients complain of blurred vision and reduced peripheral vision at presentation, but symptoms concerning peripheral vision loss tend to be vague (McDonald, 1982).The average reported duration of visual symptoms at diagnosis ranges from 6 to 24 months (Barzaghi et al., 2012;Findlay et al., 1983;Gnanalingham et al., 2005;Poon et al., 1995).
Various patterns of visual field defects have been described in patients with tumours in the pituitary region (Ogra et al., 2014;Poon et al., 1995).The precise pattern of a visual field defect depends on where the visual pathway is affected.The visual outcome following decompression varies considerably, from permanent loss of vision to complete visual recovery.The prompt diagnosis of pituitary tumours, and the detection of compression of the visual pathway, is therefore of the utmost importance to ensure optimal timing of intervention.The most important factor in predicting the post-operative prognosis of patients with pituitary tumours is the severity of the pre-operative loss of visual function (Findlay et al., 1983;Lee et al., 2018).However, the duration of symptoms at presentation cannot by itself predict post-operative prognosis of visual function (Danesh-Meyer et al., 2019).
Automated perimetry was clinically introduced in the early 1980s and offers a standardized testing method of measuring the visual field.The Humphrey Visual Field Analyser (HFA) 30-2 program tests a grid of 76 points over the central 30 degrees of the visual field in all directions from central fixation, while the 24-2 program covers a subset of the 30-2 test points and tests a smaller grid of 54 points including the central 24 degrees of the field of view and two points located at 30 degrees nasally.Since the 24-2 test pattern covers a slightly smaller area of the central visual field, it is of value to investigate whether any clinically important information is missed, compared to the 30-2 test pattern.Long test durations have previously been shown to be associated with higher variability (Hudson et al., 1994;Searle et al., 1991).Even though newer test strategies have significantly decreased the test duration (Heijl et al., 2019), examining fewer points as in the 24-2 test pattern could further shorten the test time and help reduce patient fatigue and its effect on test reliability.On the other hand, the 24-2 test pattern might exhibit lower sensitivity in detecting peripherally located visual field defects, but higher specificity particularly for false-positive defects in the superior part of the visual field often caused by droopy eyelids.The primary aim of this study was thus to determine whether the 24-2 test pattern provided comparable information to the 30-2 test pattern in detecting visual field defects in patients with tumours in the pituitary region.In addition, the location of visual field defects and correlation with visual acuity and other commonly used perimetric parameters such as mean deviation (MD) and visual field index (VFI) were also studied.

| Design
A retrospective cohort study was undertaken in patients with tumours in the pituitary region admitted for surgery at the Department of Neurosurgery, Skåne University Hospital, Lund, Sweden between the years 2000 and 2020.

| Ethical approval and consent to participate
Access to patient data was approved by the Swedish Ethical Review Authority.The research adhered to the tenets of the Declaration of Helsinki as amended in 2008.

| Inclusion criteria
Patients with tumours in the pituitary region, admitted for surgery at the Department of Neurosurgery, Skåne University Hospital, Lund, Sweden between 2000 and 2020, who had performed perimetry at Skåne University Hospital or at the patient's regional hospital within 3 months before surgery

| Exclusion criteria
Presence of other diseases that may cause visual field defects, monocular vision or neuroradiology without signs of compression of the visual pathways in the pituitary region (the optic nerves, chiasm or optic tracts).Patients were also excluded if they were not able to perform a perimetry test.A rate of false-positive responses of 15% or more was considered unreliable (Bengtsson & Heijl, 2000;Heijl et al., 2022) and these patients were also excluded.

| Visual field testing
Perimetric testing was carried out using Humphrey Field Analyzers (Carl Zeiss Meditec) models 640, 740 or 750i, with the Full Threshold, SITA Standard or SITA Fast test strategies.All visual field tests were conducted with the standard 30-2 test pattern.The results of the examination were analysed by the Statpac Single-Field Analysis implemented in the Humphrey perimeter.The Single-Field Analysis includes numerical and grayscale maps of raw threshold values of differential light sensitivities and numerical maps of deviations from agecorrected normal values with corresponding probability maps highlighting test points significantly outside normal limits at the p < 5%, <2%, <1% and <0.5% as defined by healthy subjects.The perimetric summary indices, mean deviation (MD) and visual field index (VFI), designed for staging and trend analysis of the global visual field, are also presented in the Single-Field Analysis.
As the participants already had been diagnosed with tumours and had radiologic signs of compression of the visual pathways in the pituitary region, our aim was to maintain the highest possible sensitivity in detecting any related visual field defect.Thus, we defined all significantly depressed test points (SP) in the Total Deviation probability plot flagged as statistically significantly depressed, from p < 0.05 to p < 0.005 as representing a visual field defect.The 30-2 test pattern was reduced into a corresponding 24-2 test pattern by removing the outermost ring of test points apart from two of the four nasally located points (Figure 1).The process was automated in the R programming environment RStudio version 2022.07.1 (RStudio PBC).Mean deviation (MD), visual field index (VFI) and location of visual field defects were extracted from the Statpac calculations of the 30-2 test results and hence not re-calculated based only on the 24-2 points.

| Calculation of visual field defect distribution
To analyse to what degree the visual field defects were polarized with respect to temporal-nasal or superior-inferior distribution, we calculated the percentage of the excess SP in both directions for each patient and then visualized it in a plot at the group level.The visual fields of each patient's left and right eyes were compared, regarding the total number of SP, by plotting the SP of the visual fields of the right eyes against those of the left eyes.A comparison of the number of SP in the inner 24 degrees versus the outer ring of test points (24-30 degrees) was performed by comparing the number of SP in each part (the inner 24 degrees or the outer 24-30 degrees) of the visual field against the total number of SP of the entire visual field, expressed as % SP of the total of number of test points.

| Statistical analysis
Chi-squared tests, Fisher's exact test and Pearson correlation test were performed using GraphPad Prism version 8.4.3 for Windows (GraphPad Software).

| R E SU LT S
Visual field tests obtained before surgery were evaluated for 133 patients with tumours in the pituitary region.After the application of the exclusion criteria, 79 patients (28 females and 51 males) were included in the study.Only patients with radiologically confirmed compression of the visual pathway in the pituitary region, that is the optic nerves, chiasma or optic tracts, were included in the study.Patient recruitment is illustrated in Figure 2.
The patient median age was 62 years (range 10-88), and the visual field examinations were performed in median of 7 days (range 1-91 days) before surgery.Patient characteristics are described in Figure 3.According to current visual standards, Snellen visual acuity ≥0.8 was regarded as normal visual acuity (Colenbrander, 2010).Almost two-thirds (63%) of the patients with visual field defects had a visual acuity of 0.8 or more (Figure 3a).
Evaluated from the 30-2 test results, approximately 20% of patients showed relatively small areas of visual field defects (<15 SP, Figure 3b).While most of the patients in the study had more than 15 SP, more than half of the patients only had relatively mild depression of MD (i.e.MD > −6) and equally more than half of the patients had mild depression of VFI (i.e.VFI > 80%; Figure 3c,d).
In this study, the data were first evaluated based on the results obtained with the 30-2 test pattern and then re-analysed considering only the 24-2 test points, to investigate if any clinically important information concerning visual field defects was missed when only evaluating the 24-2 part of the visual fields.In 2.5% of the analysed visual fields (4 of 158), the visual field defects were localized entirely within the area covered by the 24-2 test pattern.However, defects were observed in both the inner 24 degrees and the remaining outer ring (24-30 degrees) covering 30 degrees in 92% of the analysed visual fields (146 of 158).In 3.8% of the visual fields (6 of 158) no visual field defects were observed.Furthermore, 1.3% of the visual fields (2 of 158) had defects only in the outer ring (24-30 degrees), but signs of compression were not missed in these patients due to the detection of SP in the visual field from the other eye.One of these two patients only had one SP in the outer 30-degree ring in the eye with an otherwise normal inner 24-2 visual field, the other patient only had three SPs.Patients with very few SP in the outer ring typically also had few SP in the inner area and vice versa (Figure 4c).Thus, among the 79 patients included in the study, no visual field defects were missed when analysing the 24-2 data instead of the 30-2 data (p > 0.99, Fisher's exact test).
Mapping of the deviation in symmetry in the nasaltemporal and inferior-superior directions in participants (two visual fields per patient) showed that the majority (65%) areas of visual field defects were primarily located in the temporal region of the visual field.However, a significant minority (36%) had visual field defect that was either primarily nasal (20%) or equally distributed temporally and nasally (16%; Figure 4a).Comparing the different quadrants showed that 31% of the patients had a weighted visual field defect in the superior-temporal quadrant, 27% in the inferiortemporal quadrant, 11% superior-nasal quadrant and 7% in the inferior-nasal quadrant.The remaining 24% of patients had no weighting of visual field defects to a specific quadrant, that is evenly distributed visual field defects in all quadrants.Mapping of the symmetry in visual field defects between the left and right eye showed that although most patients had more SP in one eye, there was no difference between which eye (left or right) was more weighted when considering the patient cohort as a whole (Figure 4b).

| DI SC US SION
In the present study, no patient with radiologically confirmed tumour compression of the visual pathway in the pituitary region was perimetrically overlooked when only the 24-2 test points were considered, compared to analysis of the full 30-2 test pattern.Hence, clinical decisionmaking would not have been compromised by using the 24-2 test pattern in this patient group.In a similar study, Khoury et al. (1999) studied 187 visual fields of patients with optic nerve disease (of which 49 visual fields were diagnosed with pituitary tumours or chiasmal syndrome) using the 24-2 and 30-2 test patterns to determine if they yielded the equivalent information.They found that 95% of the visual fields in the patients with optic nerve disease were read similarly using the 24-2 and 30-2 test patterns.Of the 5% that were interpreted differently, most had an early nerve fibre bundle defect with recently diagnosed idiopathic intracranial hypertension.The authors claimed that the appropriate clinical management of these patients would not have been compromised when considering the

Excluded subjects
False positive perimetry value >15% n = 8 complete clinical picture (Khoury et al., 1999).Rowe et al. (2015) compared the HFA 30-2 and 24-2 test patterns in 50 patients with pituitary adenoma and found that 85% of the visual fields were classified as having a visual field defect using the 30-2 test pattern, whereas 80% of visual fields were classified as having a visual field defect using the 24-2 test pattern.It was not concluded if patients would have been missed when analysing the visual fields from both eyes together.In cases where the central visual defect is small or vague, or in cases where the patient only has one eye, information from the peripheral visual field may be important in determining whether the central visual field defect is truly pathological.In such rare cases, the perimetry results can be re-evaluated and if necessary, complemented with other tests.
To the best of our knowledge, there is no broad consensus on the quantitative characteristics of neurological visual field defects which makes comparisons between different studies difficult.When evaluating visual field defects using perimetry, different parameters have been used, including Total Deviation plots, Pattern Deviation plots, MD, Pattern Standard Deviation (PSD) and VFI.Each analysis has strengths and weaknesses and may be more or less appropriate when evaluating visual field defects in different diseases.The kind of analysis used should therefore be based on the suspected pathology.The MD gives an average of the total visual field defect over the complete recorded visual field, although central measuring points are weighted slightly more highly.PSD calculates the standard deviation of the values in the total deviation plot.The PSD will increase in the early stages of the disease as variation increase but decrease as a vision in larger areas of the visual field is lost, thus reducing the variation.VFI was designed to measure the linear rate of progression of visual field defect in eyes with manifest glaucoma and is much more heavily weighted towards the centre of the visual field, compared to MD.Our experience is that in clinical practice, MD and are often used to describe visual field defects in patients with pituitary tumours, although these indices were not designed to evaluate visual field defects in this context.We speculate that this use is because they are readily available and provide summary information which can be easily added to patient medical records in the short text but might also be due to a lack of detailed understanding of the perimetry algorithms.The present study indicates that MD and VFI are not reliable parameters for evaluating visual field defects due to compression, in patients with pituitary tumours, as most of the patients exhibited a large number of SP, while more than half of the patients had relatively mild depression of MD (>−6) and VFI (>80%).
In a study by Rowe et al. (2015), the criteria for abnormality using Humphrey perimetry included MD >2 dB (note from authors: probably a typo and should likely have been 'less than −2 dB') and/or a PSD >6 dB, ≥3 contiguous points at p < 5% forming a focal defect, glaucoma hemifield test outside normal limits or a combination of any two of the above.Khoury et al. considered a visual field to be abnormal when three adjacent points at p < 0.05, or two adjacent points at p < 0.01 in the pattern deviation plots were abnormal.When performing perimetry on healthy subjects, a number of points may appear to be affected.This is normally taken into account by rules defining how many neighbouring points have to be affected at a certain probability for a value to be deemed a visual field defect.The patients in the present study had already been classified as having signs of compression, or dislocation, of the visual pathways in the pituitary region (verified by radiologic signs of compression) and therefore any affected test point, at any probability level in the Total Deviation probability plot, was defined as a visual field defect.Khoury, 1999 Pattern standard deviation is evaluated in some studies (Khoury et al., 1999), although this is known to be reduced in severely affected visual fields, which may cause visual field defects to appear smaller than they are.We therefore suggest that it is better to perform analysis of the Total Deviation plots, in order not to apply any undesirable filters to the data.
Repeated measurements have been performed in several studies, where the first test is excluded in order to reduce the likelihood of an artifactual reduction in threshold sensitivity, especially in the peripheral regions, which is explained by a lack of perimetric experience (Heijl et al., 1989;Heijl & Bengtsson, 1996).In addition, previous studies have shown that the variability of the perimetric threshold generally increases in the periphery, which is explained by the periphery being more sensitive to disturbing test artefacts, such as interference from lens holders, rims of corrective lenses, eyelids or prominent brows (Heijl et al., 1987).As most patients with tumours in the pituitary region are exposed to their first visual field test, it is therefore reasonable to expect a lower degree of reliability using a 30-2 test pattern compared to a 24-2 test pattern.
Although the majority of the visual field defects were located in the temporal portion of the visual field, a significant minority of the patients in the present study exhibited most of their visual field defects in the nasal half of the visual field, which is contradictory to many textbook F I G U R E 4 Distribution of visual field defects.(a) Plot that shows the distribution of the majority of the significantly depressed test points (SP) of each visual field, where each dot represents the visual field of one eye.For each visual field, we compared the superior part versus the inferior part of the visual field by calculating the percentage difference, in number of SP, between each half.For each visual field, the same was then done in the nasal and temporal directions as well.These two values of percentage difference, that is percentage excess, were then used to plot a point, representing the visual field for each individual eye, to demonstrate to what degree and which part of the visual field had the majority of SP.For example, points plotted on the x-axis, have an equal number of SP superiorly and inferiorly, i.e. no excess in either direction, and points plotted on the y-axis equally have a 50:50 split of SP nasally and temporally.To illustrate the magnitude of defects in total, for each visual field, a colour code based on the total number of SP was used according to the scale bar.Interestingly, it was noted that although the majority of the visual fields showed a temporal polarization, a substantial proportion had nasally polarized visual fields.(b) Plot illustrating the symmetry of visual field defects comparing the left and right eyes.It was analysed by plotting the SP of the visual fields of the right eyes against those of the left eyes.Data points close to 0 on the x-axis indicate no difference in the number of SP between the patients two eyes, whereas a polarization along the x-axis indicates the degree to which one eye has more SP than the other.Most patients had more SP in one eye, but no significant difference was found between the left and right eyes on a group level (p = 0.8).(c) Plot comparing the number of statistical SP in the inner 24 degrees versus the outer ring of test points (24-30 degrees), expressed as % of SP of the total number of test points.The total number of SP is colour-coded according to the scale bar.The per cent of SP in the inner 24 degrees and in the outer ring (24-30 degrees) correlate well according to the Pearson correlation test (r = 0.92).Patients with very few SP in the outer ring typically also had few SP in the inner 24 degrees and vice versa.
examples.It also observed that 63% of the eyes had a Snellen visual acuity ≥0.8, which is often considered normal.Although alternative patterns of visual field defects have been observed by others in patients having tumours in the pituitary region (Lee et al., 2015;Ogra et al., 2014;Poon et al., 1995), these results highlight the importance of considering and describing other patterns of visual field defects, than the classical bitemporal pattern and further emphasizes that normal visual acuity is not unusual when the visual pathways are compressed.
In conclusion, the results of the present study showed that no patient, with compression of the visual pathway in the pituitary region, was missed in the detection of visual field defects, when evaluated with the 24-2 test pattern, compared to the 30-2 test pattern.However, although the majority of the visual field defects were located in the temporal visual field, as typically described for patients with tumours in the pituitary region, a large proportion of the patients showed primarily nasal visual field defects.Furthermore, the results highlight that MD and VFI are not reliable parameters for the evaluation of visual field defects due to compression in patients with pituitary tumours.In this present study, we only analysed the ability to detect any visual field defect, that is with the highest sensitivity, in the 24-2 compared to the 30-2 test pattern in a group of patients with radiologically confirmed compression of the visual pathways by a tumour in the pituitary region.In a clinical setting, it is always important to evaluate all data from patients with critical thinking, which eventually can lead to reevaluation and further investigation.

F U N DI NG I N FOR M AT ION
This study was supported by the Swedish Government Grant for Clinical Research (ALF), Skåne University Hospital (SUS) Research Grants, Skåne County Council Research Grants, the Swedish Eye Foundation, the Cronqvist Foundation and The Swedish Medical Association.

CON F L IC T OF I N T E R E ST STAT E M E N T
Boel Bengtsson has been a consultant of and is entitled to royalties from Carl Zeiss Meditec.The remaining authors declare no conflict of interest.

F
I G U R E 1 Visual field print-out.A representative example of the print-out of a Humphrey Field Analyser examination of a patient with compressive defects caused by a pituitary tumour.Greyscale maps (a) and total deviation maps (b) of right and left eyes using the 30-2 test pattern.The 30-2 test pattern results were reduced into corresponding 24-2 test patterns by removing the outermost ring of test points apart from two of the four nasally located test points (c).

F
Patient inclusion and exclusion criteria.
Patients undergoing surgery for tumors in the pituitary region at the Department of Neurosurgery, Skåne University Hospital, Sweden between 2000 and 2020, who had undergone perimetry < 3 months prior to surgery.n = 133 No radiologic sign of compression of the optic nerve n = 35 Other diseases with possible effect on visual field Glaucoma, n = 3 Amblyopia, n=1 Macular degeneration, n = 1 Possible surgical trauma to the optic nerve due to previous pituitary surgery, n = 2 Ocular hypertension, n = 1 Monocular vision, n = 2 No information on other ocular diseases, n = 1 Patients included in the study n = 79

F
I G U R E 3 Patient characteristics.The Snellen visual acuity in decimal format (a), number of significantly depressed test points (SP) at any probability level (b), mean deviation (MD) (c) and visual field index (VFI) (d) as recorded with the 30-2 test pattern.Note that 63% of the patients have a visual acuity of 0.8 or more.