Key words
non-functioning pituitary tumors - PitNets - natural history - endocrinopathies
Introduction
Non-functioning pituitary tumors (NFPTs) are non-secreting benign tumors of the
pituitary gland and are diagnosed due to endocrinopathies or mass effect causing
visual defects, or as nonfunctioning pituitary incidentalomas (NFPIs) on imaging
performed for purposes not related to the pituitary gland [1]. Transsphenoidal surgery, or a combination
of surgery and radiotherapy, is recommended as the primary treatment for NFPTs
causing mass effect symptoms [2] while surgery
of hypopituitarism is considered a relative indication [3]. However, conservative management of NFPTs
in respect to frequency of imaging and biochemical assessment of pituitary function
relies on risk estimates of tumor growth and development of new endocrinopathies,
that is, natural history of NFPTs.
A previous review regarding the natural history of NFPTs reported risk of growth of
12.5/100 person years (PYs) for macroadenomas and 3.3/100 PYs for
microadenomas [1]. However, both estimates
were associated with substantial heterogeneity and therefore do not represent
reliable estimates. Two recent cohort studies of 197 patients with NFPTs and 203
patients with NFPIs observed tumor growth in 44% of patients with NFPTs and
in 15% of patients with NFPIs during a follow-up period of 3 years [4]
[5].
Apart from tumor size, solid lesions have been associated with higher risk of tumor
growth compared to cystic lesions. A previous systematic review reported an estimate
of 2.4/100 PYs for developing new endocrinopathies during follow-up [1]. Two larger studies [4]
[5]
reported that 20 to 24% of patients had hormone deficiencies at baseline,
being more prevalent in macroadenomas, in older individuals, and in males. In these
studies, 5% experienced new endocrine dysfunction during a 3-year follow-up.
In 2017, WHO re-classified pituitary tumors based on hormonal as well as
transcription factor immunohistochemistry and also classified certain tumor types
as
having a high risk of recurrence [6].
Assessments of surgically treated patients with pituitary tumors found that
50% of tumors classified as high-risk tumors were prior to surgery
classified as non-functioning illustrating the growth potential in sub-types of
NFPTs [7]
[8] It is stressed in multiple studies [1]
[5]
[9] that the evidence regarding the natural
history of NFPTs/NFPIs is scarce and of low quality due to
heterogeneity/inconsistency, methodological limitations, and imprecision
caused by small number of events. Since the systematic review from 2011 by
Fernandez-Balsells et al. [1] a more recent
review from 2022 by Pernik et al. has been published [10], which did not include recent important
studies [11]
[12]. Both reviews report pooled estimates associated with large
heterogeneity [1]
[10] and one report estimates on risk of growth
irrespective of follow-up time [10]. While
previous review report on the risk of growth of pituitary tumors, they have failed
to report on the more clinically relevant outcome of surgical intervention.
Therefore, an updated review is called for, in order to guide management of patients
with NFPTs with respect to frequency of pituitary scan and frequency of testing for
pituitary endocrinopathies.
Objective
The objective of this systematic review was to assess the incidence of tumor growth,
new pituitary endocrinopathies, and the need for active treatment, that is, surgery
and/or radiotherapy, in conservatively treated NFPTs.
Materials and Methods
Design
A systematic review of observational studies. The protocol has been registered at
PROSPERO (CRD42020219825).
Eligibility and inclusion criteria
Eligible studies were observational studies that followed patients with NFPTs
found on CT- and/or MRI-scan and without the need for initial surgery or
radiotherapy within 3 months from baseline. Studies enrolling less than 5
patients were excluded. There were no language restrictions or restrictions on
publication date.
Outcomes
The primary outcomes were the proportion of patients with increase in tumor size,
new endocrinopathies, and the need for active treatment, that is, surgery or
radiotherapy, reported as yearly incidence per 100 person years (PYs). Secondary
outcomes were new visual defects and pituitary apoplexy. Exploratory outcomes
were the failure or recovery of specific pituitary hormonal axes and proportion
of patients with tumor shrinkage.
Identification of studies
A search of PubMed and EMBASE from inception to present date was conducted using
text words and MeSH terms as shown in Table S1. Reference lists of
included studies and relevant reviews were searched for additional eligible
studies.
Selection of studies
One investigator conducted the selection process (SD). After eliminating
duplicate studies, the remaining studies were evaluated based on title and
abstract, and potentially eligible studies were selected for full text
reading.
Data extraction
Using a standardized predefined procedure, outcomes from each study were
collected as well as the following study characteristics: country of origin,
study design, study intervention, study period, and duration of follow-up as
well as baseline patient characteristics: number of patients, gender, age,
macro- vs. microadenomas, baseline tumor size, baseline endocrine pituitary
function as failure of any axis as well as failure of a specific pituitary
hormonal axis, and baseline visual status.
Assessment of study quality
We expected to include cohort studies that did not include comparator groups and
classical bias assessment was therefore not appropriate. However, some study
characteristics may still be responsible for over- or underestimation of
outcomes and each study was characterized on six domains: Study design,
reporting of confounders, consecutive enrollment, outcome assessment, missing
data and selective outcome reporting.
Study design was described as prospective or retrospective and confounders for
tumor growth included tumor size at baseline as well as describing inclusion of
cystic tumors, while confounders for new endocrinopathies were baseline
assessment of anterior lobe hormones apart from prolactin. Patients were
considered consecutively enrolled if all patients within a defined time period,
fulfilling all inclusion criteria and none of exclusion criteria, were enrolled
in the study. Outcome assessment was considered structured in case scanning or
hormone assessment intervals were predefined the first three years from
baseline, allowing for extra scan or hormonal assessment in case of clinical
necessity. A study was classified as having missing data in case more than ten
percent of included participants were not included in the study results while
selective outcome reporting was classified in case tumor growth was not reported
or only reported if the tumor grew more than a pre-specified number of
millimeters.
Statistics
All outcomes were expressed as proportions and incidence rates per 100 person
years (PYs) with an associated 95% confidence interval (CI). The
reported number of study participants and mean duration of follow-up was used
for the calculation of total follow-up time. We expected that study populations
may differ, and a random effects model was chosen for calculation of weighted
estimates, that is, meta-analysis. Heterogeneity was assessed using the
I2-statistic, which yielded the proportion of differences in
effect size, which was caused by true differences in effect, rather than
stochastical variation. Low, moderate, and high levels of inconsistency were
defined as I
2
values of 25%, 50%, and
75%, respectively, and in case the I2>75% we
abstained from conducting a meta-analysis and only reported the range of
reported study effects. Study or patient characteristics reported in medians
were converted to means using a standard formula [13]. All analyses are presented including
all relevant studies, in case a study was considered an outlier by visual
inspection, a supplementary analysis was conducted excluding the outlier.
Publication bias was assessed by visual inspection of forest plots for the
primary outcomes and in case these were considered skewed, Egger’s test
was performed in order to qualify this observation and secondly a Duval and
Tweedie’s trim and fill procedure were conducted to estimate potential
impact of publication bias. Meta-analysis was conducted using Comprehensive
Meta-analysis 3.1, and p-values≤0.05 were considered significant.
Subgroup-analyses
Causes of heterogeneity were explored by comparing studies with low and high
proportions of macroadenomas, male subjects, and high age as well as comparing
studies with short and long follow-up time. The splitting of studies into two
groups was based on median values of each study characteristic. Variables were
also assessed for impact on outcome using meta-regression. To assess the
differences in estimates for micro- and macroadenomas we conducted direct
comparison of events in studies including both micro- and macroadenomas, which
would provide number of events obtained in the same settings and conditions
rather than compare studies limited to only one tumor size.
Deviations from the protocol
Baseline tumor extension, baseline Knosp grade, baseline co-morbidities, and time
from diagnosis to active treatment were not collected due to lack of systematic
reporting.
Results
The last search was conducted the 7th of May 2023 and as shown in [Fig. 1], [Table 1], and Table S2, we included 35 publications reporting on
30 studies [4]
[5]
[11]
[12]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]
[39]. The included studies
represent data from 1957 patients with NFPTs who were initially treated
conservatively and followed for a mean of 4.0 (SD 1.5) years. Two of the included
studies followed a total of 69 patients with Multiple Endocrine Neoplasia, type 1
(MEN1) [28]
[37] and two studies followed 99 children [12]
[39]. The mean of study
participants mean age was 48.2 years (SD 15), while the mean proportion of male
subjects was 46 (SD 15) percent. At baseline, the mean proportion of macroadenomas
in each study was 55 (SD 36) percent, the mean proportion of participants with any
pituitary deficiency was 26 (SD 25) percent, and 12 (SD 18) percent had visual
deficiencies.
Fig. 1 Identification and selection of included studies.
Table 1 Baseline characteristics of studies assessing
progression of conservatively treated non-functioning pituitary
tumors.
|
Author [Ref], Year of publ.
|
n (mean age)
|
Male (%)
|
MacroA (%)
|
Endocrinop. (E) Visu. Defect (V)
|
Scan
|
Years of follow-up, mean (range)
|
|
Reincke [14], 1990
|
14 NFPI (41)
|
3/14 (21)
|
7/14 (50)
|
E: 3/14 V: 0/14
|
Yearly
|
3.2 (0.9 to 8)
|
|
Donovan [15], 1995
|
31 NFPI (35)
|
11/31 (35)
|
16/31 (52)
|
E: 0/31 V: 0/31
|
After 6 mo. – then yearly
|
6.4 (3 to 11)
|
|
Nishizawa [16], 1998
|
28 NFPI (63)
|
13/28 (46)
|
28/28 (100)
|
E: 7/28 V: 0/31
|
Yearly
|
5.6 (0.5 to 10)
|
|
Feldkamp [17], 1999
|
50 NFPI (NR)
|
NR
|
19/50 (38)
|
E: NR V: 0/50
|
After 3 mo. – then yearly
|
2.7
|
|
Igarashi [18], 1999
|
23 NFPT (47)
|
10/23 (44)
|
22/23 (96)
|
E: 1/23 V: 8/23
|
NR
|
5.1 (1.5 to 11.6)
|
|
Oyama [19], 2005
|
289 NFPI (49)
|
130/289 (45)
|
NR
|
E: NR V: 0/289
|
After 3 mo. – then NR
|
2.3 (0.5 to 14.5)
|
|
Vilar [20], 2005
|
12 NFPI (NR)
|
NR
|
4/12 (33)
|
E: NR V: NR
|
NR
|
2.8
|
|
Arita [21], 2006
|
42 NFPI (61)
|
18/42 (43)
|
37/42 (88)
|
E: NR V: 0/42
|
After 6 mo. – then yearly
|
5.2 (0.9 to 14.0)
|
|
Dekkers [22], 2007
|
28 NFPT (55)
|
15/28 (54)
|
28/28 (100)
|
E: 20/28 V: 13/28
|
Yearly or every 2 year
|
7.1
|
|
Karavitaki [23], 2007
|
40 NFPT (52)
|
18/40 (45)
|
24/40 (60)
|
E: 11/40 V: 5/40
|
Every 1–2 years
|
3.5 (0.7 to 10.7)
|
|
Carsote [24], 2009
|
69 NFPT (NR)
|
NR
|
0/69 (0)
|
E: 0/69 V: NR
|
NR
|
2.5 (0.8 to 8)
|
|
Cury [25], 2009
|
13 NFPT (NR)
|
NR
|
NR
|
E: 9/13 V: 7/13
|
NR
|
6.9
|
|
Ryu [26], 2010
|
6 NFPT (66)
|
5/6 (83)
|
6/6 (100)
|
E: NR V: 1/6
|
Every 6–12 mo.
|
3.4
|
|
Anagnostis [27], 2011
|
23 NFPT (NR)
|
NR
|
6/23 (26)
|
E: NR V: NR
|
At least yearly
|
4.6
|
|
de Laat [28], 2015
|
45 NFPT (NR)
|
NR
|
6/45 (13)
|
E: NR V: NR
|
NC
|
5.5
|
|
Karamouzis [29], 2015
|
33 NFPT (60)
|
21/33 (64)
|
27/33 (82)
|
E: NR V: NR
|
At least yearly
|
4.5 (1 to12)
|
|
Sam [30], 2015
|
66 NFPT (41)
|
28/66 (42)
|
47/66 (71)
|
E: 40/66 V: 5/66
|
Every 6–12 mo.
|
4.3 (1 to14.7)
|
|
Vargas [31], 2015
|
19 NFPT (NR)
|
NR
|
NR
|
E: NR V: NR
|
NR
|
NR
|
|
Imran [32], 2016
|
99 NFPI (NC)
|
NR
|
NR
|
E: NR V: NR
|
Every 6–12 mo.
|
3.0
|
|
Lenders [33], 2016
|
50 NFPT (49)
|
15/50 (30)
|
23/50 (46)
|
E: 4/50 V: 0/50
|
NR
|
3.0 (0.5 to 6.6)
|
|
Iglesias [34], 2017
|
26 NFPI (NR)
|
NR
|
9/26 (35)
|
NR
|
Yearly or more
|
1.3
|
|
Kim [4], 2018
|
197 NFPT (53)
|
96/197 (49)
|
159/197 (81)
|
E: 40/197 V: 0/197
|
Yearly for 2 years
|
3.1
|
|
Levy [35], 2018
|
65 NFPT (68)
|
37/65 (57)
|
65/65 (100)
|
E: NR V: 23/65
|
After 6 mo. then clinics
|
5.0
|
|
Stalldecker [36], 2019
|
11 NFPT (>65)
|
NR
|
NR
|
E: NR V: 1/11
|
NR
|
3.6
|
|
Thaker [12], 2019
|
44 NFPT (14)
|
8/44 (18)
|
3/44 (6.8)
|
E: 7/44 V: 0/44
|
Yearly
|
4.5
|
|
Wu [37], 2019
|
24 NFPT (44)
|
12/24 (50)
|
NR
|
E: NR V: NR
|
NR
|
2.9 (0.5 to 7.3)
|
|
Hwang [38], 2020
|
81 NFPT (58)
|
48/81 (59)
|
81/81 (100)
|
E: 0/81 V: 0/81
|
Every 2 year
|
5.5
|
|
Tresoldi [5], 2020
|
203 NFPI (50)
|
NR
|
71/203 (35)
|
E: NR V: 63/177
|
After 6–12 mo.
|
3
|
|
Han [11], 2022
|
271 NFPT (NR)
|
NR
|
0/271 (0)
|
E: NR V: 0/271
|
NR
|
2.4 (0 to 12.8)
|
|
Borghammar [39], 2023
|
55 NFPT (12)
|
24/55 (44)
|
0/55 (0)
|
E: 22/55 V: 0/55
|
NR
|
3 (0.3 to 15.8)
|
NR: Not reported or unclear; MacroA: Macroadenoma; mo.: Months; NFPT:
Non-functioning pituitary tumors; NFPI: Non-functioning pituitary
incidentalomas.
Quality of studies
As shown in Table S4, 1 of 30 studies was classified as a prospective study,
information on tumor size at baseline as well as proportion of cystic tumors
were reported in 6 of 30 studies, while baseline endocrinopathies were reported
in 17 of 30 studies. Missing data, regarding tumor size and development of new
endocrinopathies, was observed in 11 and 15 of 30 studies.
Primary outcomes
Tumor growth
During follow-up, tumor growth was observed in 422/1914
(22.0%) patients, however, assessed as events per 100 PYs the
between study heterogeneity was high
(I2
=89%) ranging from a risk of 0.5 to
14.2/100 PYs as shown in [Table
2] and [Table 3]. Exploring
causes of heterogeneity, we found that studies with a high proportion of
macroadenomas reported a higher risk of tumor growth while mean age,
proportion of males, and mean follow-up time did not explain heterogeneity
(Table S5) and further exploration of heterogeneity using
meta-regression confirmed these findings while suggesting that increasing
age may be associated with tumor growth: proportion of macro-adenomas
(p=0.006), mean age (p=0.03), proportion
of males (p=0.13), and the mean follow-up time
(p=0.72).
Table 2 Events during follow-up of conservatively
treated non-functioning pituitary tumors.
|
Study [ref], Year of publ.
|
Growth All tumors
|
Growth MacroA
|
Surgery All tumors
|
Surgery MacroA
|
New endocrinop. All tumors
|
New endocrinop. MacroA
|
Visual deterioration All tumors
|
|
Events/total
|
|
|
|
|
|
|
|
Reincke [14],
1990
|
3/14
|
2/7
|
0/14
|
0/7
|
1/14
|
1/7
|
0/14
|
|
Donovan [15],
1995
|
4/31
|
4/16
|
1/31
|
1/16
|
0/31
|
0/16
|
1/31
|
|
Nishizawa [16],
1998
|
1/28
|
1/28
|
2/28
|
2/28
|
NR
|
NR
|
2/28
|
|
Feldkamp [17],
1999
|
6/50
|
5/19
|
NR
|
NR
|
NR
|
NR
|
0/50
|
|
Igarashi [18],
1999
|
9/23
|
9/22
|
9/23
|
9/22
|
NR
|
NR
|
9/23
|
|
Oyama [19],
2005
|
30/289
|
NR
|
11/289
|
NR
|
NR
|
NR
|
NR
|
|
Vilar [20],
2005
|
1/12
|
1/4
|
2/12
|
1/4
|
NR
|
NR
|
NR
|
|
Arita [21],
2006
|
21/42
|
19/37
|
12/42
|
9/37
|
NR
|
NR
|
NR
|
|
Dekkers [22],
2007
|
14/28
|
14/28
|
6/28
|
6/28
|
3/28
|
3/28
|
9/28
|
|
Karavitaki [23],
2007
|
14/40
|
12/24
|
8/40
|
8/24
|
NR
|
NR
|
8/40
|
|
Carsote [24],
2009
|
19/69
|
0/0
|
NR
|
NR
|
NR
|
NR
|
NR
|
|
Cury [25], 2009
|
0/13
|
NR
|
0/13
|
0/13
|
0/13
|
NR
|
0/13
|
|
Ryu [26], 2010
|
4/6
|
4/6
|
1/6
|
1/6
|
NR
|
NR
|
2/6
|
|
Anagnostis [27],
2011
|
1/23
|
1/6
|
NR
|
NR
|
NR
|
NR
|
NR
|
|
de Laat [28],
2015
|
3/45
|
0/6
|
1/45
|
1/6
|
0/45
|
0/6
|
NR
|
|
Karamouzis [29],
2015
|
5/33
|
4/27
|
NR
|
NR
|
NR
|
NR
|
NR
|
|
Sam [30], 2015
|
38/66
|
28/47
|
9/66
|
9/47
|
NR
|
NR
|
9/66
|
|
Vargas [31],
2015
|
6/19
|
NR
|
NR
|
NR
|
NR
|
NR
|
NR
|
|
Imran [32], 2016
|
20/99
|
NR
|
NR
|
NR
|
1/77
|
NR
|
NR
|
|
Lenders [33],
2016
|
11/50
|
9/23
|
6/50
|
5/23
|
2/50
|
2/23
|
1/50
|
|
Iglesias [34], 2016
|
1/26
|
0/9
|
NR
|
NR
|
NR
|
NR
|
NR
|
|
Kim [4], 2018
|
87/197
|
NR
|
8/197
|
8/159
|
8/197
|
8/159
|
5/197
|
|
Levy [35], 2018
|
23/65
|
23/65
|
13/65
|
13/65
|
NR
|
NR
|
NR
|
|
Stalldecker [36],
2019
|
1/11
|
NR
|
NR
|
NR
|
0/11
|
NR
|
1/11
|
|
Thaker [12], 2019
|
1/44
|
0/3
|
0/44
|
0/3
|
NR
|
NR
|
0/44
|
|
Wu [37], 2019
|
NR
|
NR
|
NR
|
NR
|
NR
|
NR
|
NR
|
|
Hwang [38], 2020
|
51/81
|
51/81
|
16/81
|
16/81
|
10/81
|
10/81
|
14/81
|
|
Tresoldi [5],
2020
|
31/203
|
19/71
|
13/203
|
12/71
|
10/194
|
3/62
|
9/39
|
|
Han [11], 2022
|
22/271
|
0/0
|
0/271
|
0/0
|
5/266
|
0/0
|
0/0
|
|
Borghammar [40],
2022
|
0/55
|
0/0
|
0/55
|
0/0
|
NR
|
NR
|
NR
|
NR: Not reported or unclear; MacroA: Macroadenoma; Endocrinopat:
Endocrinopathies.
Table 3 Incidence of events during follow-up in
conservatively treated non-functioning pituitary
tumors.
|
Events/total (%)
|
Studies n
|
Range of events//100 PYs
|
Incidence/100 PYs1 (95% CI;
I2)
|
|
Tumor growth
|
|
Overall
|
421/1914 (22.0)
|
28
|
0 to 14.2
|
I2
=90%
|
|
Macroadenomas
|
201/510 (39.4)
|
19
|
0 to 19.5
|
I2
=82%
|
|
Microadenomas
|
57/675 (8.4)
|
15
|
0 to 12.2 (0 to 7.7)2
|
1.8 (0.9 to 2.8; 58%)
|
|
New endocrinopathies
|
|
Overall
|
40/1007 (4.0)
|
12
|
0 to 2.3
|
0.9 (0.5 to 1.2; 35%)
|
|
Macroadenomas
|
27/382 (7.1)
|
9
|
0 to 4.5
|
1.5 (0.9 to 2.1; 0%)
|
|
Microadenomas
|
11/277 (4.0)
|
6
|
0 to 3.8
|
0.7 (0.0 to 1.6; 37%)
|
|
Need for active treatment
|
|
Overall
|
119/1603 (7.4)
|
21
|
0 to 7.7
|
I2
=80%
|
|
Macroadenomas
|
101/640 (15.8)
|
18
|
0 to 8.9
|
3.1 (2.1 to 4.1; 52%)
|
|
Microadenomas
|
19/673 (2.8)
|
13
|
0 to 11.5 (0 to 4.5)3
|
0.5 (0.1 to 0.9; 37%)
|
|
Recovery of endocrinopathy
|
|
Overall
|
14/264 (5.3)
|
3
|
0.5 to 3.5
|
1.2 (0.1 to 2.3; 60%)
|
|
New visual defects
|
|
Overall
|
70/992 (7.1)
|
18
|
0 to 9.8
|
I2
=78%
|
|
Macroadenomas
|
69/517 (13.3)
|
13
|
0 to 10
|
2.6 (1.5 to 3.8; 65%)
|
|
Microadenomas
|
1/403 (0.2)
|
8
|
0 to 3.7
|
0.1 (0.0 to 0.3; 0%)
|
|
Pituitary apoplexy
|
|
Overall
|
21/826 (2.5%)
|
10
|
0 to 3.4
|
0.4 (0.1 to 0.7; 23%)
|
|
Tumor shrinkage
|
|
Overall
|
248/1619 (15.3%)
|
20
|
1.3 to 19.7 (1.3 to 7.8)4
|
I2
=80%
|
1 Incidence per 100 person years is based on a random
effects meta-analysis. Estimate is only calculated if
I2≤75%. 2 One outlier removed,
Sam et al. [30], changing the
estimate to 1.6/100 PYs (95% CI. 0.8 to 2.4;
I2
=49%) 3 One
outlier removed, Arita et al. [21], changing the estimate to 0.4/100 PYs
(95% Cl. 0.1 to 0.8;
I2
=32%) 4 One outlier
removed, Carsote et al. [24],
changing the estimate to 3.7/100 PYs (95% Cl. 2.6 to
4.7; I2
=74%) PYs: Person
years.
As shown in [Table 3], high
heterogeneity (82%) was still present limiting the analysis to
macroadenomas, while the pooled estimate of growth in microadenomas was
1.8/100 PYs (95% CI; 0.9 to 2.8;
I2
=58%). Excluding one outlier [30] reporting a very high risk of
growth in microadenomas, the risk of growth was reduced to 1.6/100
PYs (95% CI; 0.8 to 2.4; I2 =49%).
Analysis comparing the risk of growth in studies including both micro- and
macroadenomas showed significant lower risk of growth in macroadenomas
compared to microadenomas (p=0.002), as shown in Table
S5.
New endocrinopathies
A total of 40/1007 (4.0%) patients developed a new pituitary
endocrinopathy during follow-up, corresponding to a risk estimate of
0.9/100 PYs (95% CI; 0.5 to 1.2;
I2
=35%). Subgroup analysis found that
studies with >39% males had a higher risk of developing new
endocrinopathies during follow-up, which was not confirmed using
meta-regression analysis. As shown in Table S6, analysis of the other
pre-planned subgroups did not explain the observed heterogeneity. There were
no significant associations between the explored subgroups and risk of
events assessed by meta-regression: mean age (p=0.66),
proportion of males (p=0.19), proportion of macroadenomas
(p=0.08), or mean follow-up (p=0.76).
In macroadenomas, the risk of developing new endocrinopathies was 1.5
(95% CI; 0.9 to 2.1; I2
=0%)
compared to 0.7 (95% CI; 0.0 to 1.6;
I2
=37%). Limiting the analysis to studies
including both micro- and macroadenomas the risk of new endocrinopathies was
0.9/100 PYs in microadenomas compared to 2.1/100 PYs, which
was not significantly different (p=0.15).
Active treatment
During follow-up the total number of patients with the need for active
treatment was 119/1603 (7.4%), ranging from 0 to
7.7/100 PYs with large heterogeneity,
I2
=80%. Among all patients receiving
active treatment only one patient received RT [18], while the rest underwent pituitary
surgery. As shown in Table S7, none of subgroup analyses explained
any of the observed heterogeneity. Exploring the influence of subgroups
using meta-regression supported the above findings: mean age
(p=0.02), proportion of males (p=0.95),
proportion of macroadenomas (p=0.05), and mean follow-up
(p=0.78). A separate analysis of macroadenomas found the
risk of surgery to be 3.1/100 PYs (2.1 to 4.1;
I2
=52%) compared to 0.5 PYs
(95% CI; 2.1 to 4.1; I2
=37%) in
microadenomas. This was confirmed in a formal analysis assessing studies
including both micro- as well as macroadenomas finding the risk of active
treatment significantly higher in macroadenomas
(p=0.001).
Publication bias
Inspection of the funnel plots suggested that studies reporting low incidence
risk was missing for all primary outcomes, a finding which was supported by
Egger’s test (all p<0.04). Using Duval and Tweedie’s
trim and fill procedure the overall estimate was reduced by 0.1, 0.3, and
0.5/100 PYs for growth, new endocrinopathies, and active treatment.
Year of publication did not affect any of the primary outcomes in
meta-regression analysis (all p-values>0.25).
Secondary outcomes
During follow-up the total number of patients with new visual deficiencies
was 70/992 (7.1%), with the lowest and highest reported
incidence being 0.0 to 9.8/100 PYs with an associated I2
of 78% and as shown in [Table
3], a new visual defect only occurred in 1 of 403 patients with
microadenomas. Pituitary apoplexies occurred in 21/826
(2.5%) patients corresponding to an incidence of 0.4 100/PYs
with an associated I2 of 23%.
Exploratory outcomes
Tumor shrinkage was observed with an incidence between 1.3 to
19.7/100 PYs (I2
=80%), however,
removing one outlier [24] the
incidence of tumor shrinkage across studies was estimated to 3.7/100
PYs as shown in [Table 3], while the
overall risk of recovery of any pituitary function was estimated to
1.2/100 PYs.
Failure of a specific hormonal axis was only reported in four studies
(Table S3) with risk of ACTH-failure having the highest risk of
0.8/100 PYs as shown in Table S8. No studies reported on the
risk of diabetes insipidus.
Discussion
This systematic review, including 1957 conservatively treated NFPTs followed for a
mean of 4.0 years, found that the reported risk of tumor growth and surgical
interventions was higher in macroadenomas compared to microadenomas. While this
difference was also observed in the risk of new endocrinopathies it did not reach
statistical significance. The important patient outcomes of surgery, new
endocrinopathies, or deterioration of visual function was all below 1.0/100
PYs in patients with microadenomas.
Strengths and limitations
A major strength of this systematic review is the rigorous bibliographical search
strategy, a published protocol, bias assessment of included studies, pre-defined
subgroup analysis, and a transparent reporting. Also, separate analysis of
micro- and macroadenomas across several outcomes provides strong and clinical
meaningful estimates for these separate entities. However, funnel plot of
primary outcomes suggested that studies with lower incidence rates was missing
and that the pooled estimates may be lower. In case the funnel plots represent
true bias, the trim and fill procedure did not change the estimated incidences
in a magnitude that changed the interpretation of current results. The
calculation of observed patients’ years was based on mean follow-up
times opposed to individual follow-up time, which may distort the estimates in
either direction. Frequency of scans, definition of growth, testing of pituitary
function and indications for pituitary surgery may differ from study to study,
which all may contribute to heterogeneity of the observed results.
Primary outcomes
The risk of tumor growth overall and for macroadenomas was associated with high
heterogeneity with a range of reported incidence for growth of macroadenomas
between 0 and 19.5/100 PYs. A previous review found a risk of growth of
macroadenomas of 12.5/100 PYs, an estimate associated with an
I2
=99% [1]. In the current study, only 5 of 20 studies report incidence rates
beyond 12.5 and it likely that the true incidence rates are lower. In a
sub-group analysis of studies including both micro- and macroadenomas the
estimate for growth in macroadenomas was 7.0/100 PYs, with an associated
I2 of 66%. Using the 2017 WHO classification of pituitary
adenomas including immunohistochemistry and transcription factor analysis,
Lenders et al. classified 111 NFPTs into more than 10 different subtypes [8], and thus highlighting that NFPTs are
likely to have a different prognosis and growth pattern contributing to the
observed heterogeneity. The risk of growth in microadenomas was estimated to
1.8/100 PYs among 675 patients in the current study. Removing one study
[30], the estimate was reduced to
1.6/100 PYs providing a solid estimate based on individual patient data
and low to moderate heterogeneity. The study by Sam et al. [30] did exclude patients who had less than
a year of follow-up, which may imply that smaller microadenomas were excluded
thus enriching the cohort with more severe cases. Compared to the systematic
review performed by Fernandez-Balsell et al. in 2010 [1], the current findings suggest lower
incidence rates of growth for both micro- and macroadenomas. Removing one
outlier [30], the risk of tumor regression
was estimated to 3.7/100 PYs ([Table
3]), which may reflect either that the most prevalent definition of a
significant tumor change of 2 mm may be too small causing uncertain
estimates or that tumor regression is a common event.
The overall risk of developing a new endocrinopathy was estimated to
0.9/100 PYs, an estimate associated with an I2 of
35%. Separate analysis of micro- and macroadenomas suggested that new
endocrinopathies occurred twice as frequently in macroadenomas compared to
microadenomas with incidence rates of 1.5/100 PYs and 0.7/100
PYs, respectively. However, in the subgroup analysis of studies reporting on
both micro- and macroadenomas, the difference was not significant. These
estimates are much lower than previously reported by Fernandez-Balsell reporting
an incidence of 11.9/100 PYs and 4.0/100 PYs for macro- and
microadenomas [1]. Theses discrepancies
were not explained by time of publication, however, only three of 12 studies
included on this outcome were published prior to 2011 and included in the work
by Fernandez-Balsell et al. [1] and the
observed variations may be explained partly by chance.
The risk of undergoing active treatment, that is, surgery, was estimated to
3.1/100 PYs in macroadenomas and 0.4/100 PYs in microadenomas,
both estimates were associated with small to medium heterogeneity. This is an
obvious important outcome and to our knowledge this is the first time for a
systematic review to report on this outcome. As described previously [3] surgery may be performed due to chiasma
pressure or for recovery of pituitary function, with the later not being
practiced all places, which may contribute to the differences in estimates.
Secondary outcomes
Tumor shrinkage was reported with an incidence of 3.7/100 PYs and in
comparison, the risk of growth was 1.8/100 PYs in microadenomas.
Pituitary apoplexies are often associated with shrinkage of tumor size; however,
this was only observed in 0.4/100 PYs and is unlikely to explain the
high incidence of tumor shrinkage. The definition of shrinkage was rarely
reported, and it is possible that studies using as little as 1 or 2 mm
as cut offs for growth or shrinkage may report higher incidence of change in
tumor size due to misreading.
Perspectives
Guidelines from the Endocrine Society on pituitary incidentalomas recommend that
microadenomas should have a repeat MRI-scan after 1 year, and if the
incidentaloma does not change in size, then every 1–2 years the
following three years and subsequently with gradually less frequency [40]. With stable size of the microadenoma,
regular endocrine assessment is not recommended, reflecting the association
between tumor size and risk of pituitary endocrine failure. The current findings
strongly support differentiated follow-up of patients with microadenomas
compared to macroadenomas. Clinically relevant growth in microadenomas is rare
and resources should primarily be allocated to follow-up of macroadenomas with
higher risk of growth, surgery and failure of endocrine function. Predicting
growth behavior in NFPTs prior to pathological classification remains primarily
based on tumor size and tumor behavior in terms of aggressiveness and
invasiveness. The current systematic review identifies a low risk of growth, low
risk of new endocrinopathies and low risk of surgical interventions in patients
with microadenomas and we suggest that patients with normal endocrine function
at baseline and no signs of tumor growth at a two- or three-year MRI follow-up
could end any further assessment of pituitary function or size. This would apply
to well informed patients that would be able to respond to symptoms associated
with tumor growth, that is, visual disturbances or failure of pituitary
function, while less able patients may remain under some sort of endocrine
observation.
Conclusion
The risk of growth in NFPTs is lower than previously reported. In particular, the
low
risk of growth, new endocrine failure, or need for surgery is low in microadenomas
calling for shorter duration of follow-up than suggested by current guidelines.
Future studies with long follow-up and detailed description of endocrine function
and tumor characteristics are currently warranted in order to improve
pre-surgical/pre-pathological guidance on the risk of clinically relevant
outcomes, especially for macroadenomas.
