Pheochromocytomas and paragangliomas in von Hippel-Lindau disease: not a needle in a haystack.

OBJECTIVE: Pheochromocytomas are a hallmark feature of von Hippel-Lindau disease (vHL). To our knowledge, this is the first systematic review with meta-analysis evaluating the frequency of pheochromocytomas and/or paragangliomas (PPGLs) in patients with vHL, as well as among patients with different vHL subtypes.
DESIGN: Systematic review with meta-analysis.
METHODS: We searched on MEDLINE, Scopus, and Web of Science. We included primary studies assessing participants with vHL and reporting on the frequency of PPGL. We performed random-effects meta-analysis to quantitatively assess the frequency of PPGL, followed by meta-regression and subgroup analysis. Risk of bias analysis was performed to assess primary studies' methodological quality.
RESULTS: We included 80 primary studies. In 4263 patients with vHL, the pooled frequency of PPGL was 19.4% (95% CI = 15.9-23.6%, I2 = 86.1%). The frequency increased to 60.0% in patients with vHL type 2 (95% CI = 53.4-66.3%, I2 = 54.6%) and was determined to be of 58.2% in patients with vHL type 2A (95% CI = 49.7-66.3%, I2 = 36.2%), compared to 49.8% in vHL type 2B (95% CI = 39.9-59.7%, I2 = 42.7%), and 84.1% in vHL type 2C (95% CI = 75.1-93.1%, I2 = 0%). In meta-regression analysis, more recent studies were associated with a higher frequency of PPGL. All studies had at least one internal validity item classified as 'high risk of bias,' with 13% studies having low risk of bias in all external validity items.
CONCLUSIONS: PPGLs are a common manifestation of vHL. Despite methodological limitations and differences across primary studies, our results point to the importance of PPGL screening in patients with vHL.

Introduction

Von Hippel–Lindau disease (vHL) is a rare hereditary autosomal dominant disorder resulting from the presence of pathogenic variants on the VHL gene (1), with a point prevalence estimated (as per the first nationwide study on the subject) at 1:46,900 individuals, a birth incidence at 1:27,300 live births, and an overall penetrance of 87% at 60 years of age (2). Carriers of disease-causing VHL germline variants are at an increased risk of developing benign and malignant neoplasms, such as hemangioblastomas and pheochromocytomas. Pheochromocytomas and paragangliomas (PPGLs), arising either through germline VHL mutations in a syndromic setting or via somatic VHL mutations sporadically, associate with the pseudo-hypoxic mRNA expressional cluster 1 in which an increased risk of metastatic events is seen compared to tumors arising within the kinase-associated cluster 2, containing more biologically indolent tumors (3, 4). Pheochromocytomas are more frequently diagnosed during the third decade of life although they may also be the presenting feature in the pediatric setting (2, 5). Pheochromocytoma is such a hallmark feature of vHL that its absence or presence, respectively, underlies the phenotypic classification in type 1, which is typically associated with protein-truncating mutations, or vHL type 2, which mainly associates with missense mutations (6). vHL type 2 is further subdivided into types 2A (low risk of renal cell carcinoma), 2B (high risk of renal cell carcinoma), and 2C (only pheochromocytoma without other vHL manifestations) (6, 7). In the clinical setting, even though surveillance protocols are generally similar for all vHL patients, this classification is particularly useful for appropriate intrafamilial risk management and prognosis assessment since the risk of developing pheochromocytomas in vHL-affected family members increases once the history of pheochromocytoma in the family is reported and the family is considered of type 2 vHL.

An Italian nationwide prospective study assessing individuals with vHL found that 30% of participants developed pheochromocytoma, resulting in a cumulative incidence of adrenal disability of 11% after surgery, requiring adrenal substitutive therapy. Those patients underwent surgery for pheochromocytoma at a mean age of 27 years, similar to those who underwent surgery for hemangioblastoma (28 years) and earlier than those who underwent surgery for renal and pancreatic neoplasms (37 and 35 years old, respectively) (8). Therefore, international clinical practice guidelines support early detection of pheochromocytomas in vHL patients, as it may allow for more advanced surgical techniques – such as cortical-sparing and laparoscopic procedures – to be performed, resulting in lower risk of recurrence and maintenance of corticosteroid independence (9, 10).

However, notwithstanding the burden of pheochromocytoma in vHL and despite the benefits of early and lifetime screening for this neoplasm in patients with vHL (11), the prevalence of pheochromocytoma in patients with vHL has not been systematically ascertained. Several studies have reported the frequency of pheochromocytoma in cohorts of patients with vHL, but with relevant disparities, mirroring their differences in study design, methodological quality, and sample size. Thus, this systematic review aims to determine the frequency of pheochromocytoma and/or PPGL in patients with vHL (and in patients with each vHL subtype), as well as to identify variables potentially explaining across-study differences on the frequency of PPGL.

Materials and methods

This study corresponds to a systematic review with meta-analysis following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (12).

Eligibility criteria

To determine the frequency of PPGL in vHL, we included cohorts of patients with vHL – either fulfilling clinical diagnostic criteria of vHL or carrying VHL germline pathogenic mutations – which have reported data on the number of patients/mutation carriers with PPGL. In order to determine the frequency of PPGL in each vHL subtype, we also included family reports that have reported data on the number of patients/mutation carriers with vHL of each subtype with PPGL. To assess the frequency of PPGL in vHL type 2C, we included only studies that have performed molecular screening of asymptomatic family members (once PPGL is the only phenotypic feature in vHL type 2C, all patients clinically diagnosed with vHL type 2C have a history of PPGL; therefore, we included only primary studies that performed genetic screening of asymptomatic family members so that we were able to assess the frequency of PPGL in all family members with the vHL type 2C mutation). We excluded case–control studies, case series of pheochromocytoma/PPGL patients, and studies in which patient selection was based on specific phenotypic manifestations (e.g. case series of vHL patients with hemangioblastomas).

Search strategy and study selection

Studies were identified in February 2020 through a search conducted on three electronic bibliographic databases (MEDLINE/PubMed, Web of Science, and Scopus) using the queries displayed in Supplementary Table 1 (see section on supplementary materials given at the end of this article). No language or date restrictions were applied.

Following duplicate removal, screening of the titles and abstracts was independently conducted by two researchers. The full texts of selected studies were then independently read by two researchers, who decided on their final inclusion. For publications in which complete or partial overlap of participants with other studies occurred, we presented only information from the publication with more participants, eventually complementing with relevant information from other publications on covariates of interest.

Data extraction

Two researchers independently extracted relevant data from included studies regarding their year of publication, country, sampling method, method of vHL diagnosis (clinical and/or molecular), number of vHL patients/mutation carriers, participants’ gender and age distributions, number of asymptomatic carriers of VHL pathogenic mutations, number of patients with positive family history of vHL disease, and number of patients with PPGL (including the number of patients with bilateral adrenal presentation). Our outcome variable consisted of the number of patients with PPGL (instead of the number of patients with pheochromocytoma), as many studies did not differentiate between adrenal pheochromocytoma and PPGL/extra-adrenal pheochromocytoma (13). Nevertheless, we also collected separate data regarding the frequency of pheochromocytoma and PPGL whenever these entities were adequately distinguished. We also retrieved information concerning the number of vHL patients/mutation carriers with each vHL phenotypic subtype (vHL type 2A, vHL type 2B, vHL type 2C) and the number of patients with each vHL subtype who had PPGL.

For data collection, we used a purpose-built form developed after a pilot version. Authors were contacted to provide relevant missing information. In cases of disagreement regarding study selection or data extraction, a consensus was achieved.

Quality assessment

Study quality was independently assessed by two researchers according to an adapted version of the Hoy et al. judgments on the risk of bias of prevalence studies (14). Three risk of bias items were excluded as they did not adequately apply to our study, namely that regarding data provenance (directly from subjects vs from a proxy) on account of the type of data we were retrieving, that concerning reliability and validity of the study instrument used to measure the parameter of interest, and that regarding the suitability of the length of the shortest prevalence period (as we were not assessing a specific prevalence period).

Quantitative synthesis

We performed random-effects meta-analyses to estimate the pooled frequency of PPGL among patients with vHL, as well as among patients with each vHL subtype. In addition, we computed the meta-analytical frequency of patients with PPGL who presented with bilateral neoplasms or who were of the male sex. To account for possible bias resulting from studies assessing small samples (with the possibility of most patients being from a small number of families), we performed sensitivity analyses restricted to studies with more than 25 participants.

Meta-analysis was performed based on the restricted maximum likelihood method, using logit-transformed proportions. Pooled results were back-transformed to their original scale to facilitate their interpretation. Heterogeneity was assessed using the I2 statistic and Cochran Q test P  value – an I2 > 50% and a P  value < 0.10 were considered to represent substantial heterogeneity. In the presence of substantial heterogeneity, we performed meta-regression and subgroup analyses in order to identify variables possibly explaining across-study differences. Covariates tested include the publication year, sample size, percentage of male participants, mean participants’ age, region (Europe, America, and Asia and Pacific), sampling method (consecutive, convenience, or not specified), and vHL diagnosis method (clinical criteria only or including genetic testing/not specified). All statistical analyses were performed using metafor package of software R (version 4.0.0.).

Results

Search results

Our search yielded 2557 publications, of which 906 were duplicates. One thousand four hundred ten studies were excluded on the basis of application of eligibility criteria (Fig. 1). Fifty-five records were excluded due to partial or complete overlap of participants with included publications. We could not obtain the full text of 13 publications despite contacting corresponding authors (Supplementary Table 2). Seven of these articles were published in a language other than English, and six were published before 2000.

Figure 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram illustrating study selection process.

A total of 80 studies were included in this systematic review. Of these, 45 cohort studies were used in the meta-analytical quantitative synthesis of the frequency of PPGL among participants with vHL (8, 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, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58). On the other hand, 52 studies were used in the frequency analysis of PPGL in vHL type 2 (16, 17, 21, 25, 26, 28, 29, 35, 36, 40, 44, 45, 47, 48, 53, 55, 56, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93), of which 24 corresponded to cohort studies (16, 17, 21, 25, 26, 28, 29, 35, 36, 40, 44, 45, 47, 48, 53, 55, 56, 62, 63, 64, 66, 77, 78, 79) and 28 to family reports (59, 60, 61, 65, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93). In total, 17 studies were used in both analyses (16, 17, 21, 25, 26, 28, 29, 35, 36, 40, 44, 45, 47, 48, 53, 55, 56).

Frequency of PPGL among participants with vHL

The meta-analysis assessing the frequency of PPGL comprised 45 studies with a total of 4263 participants with vHL (8, 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, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58). Most of these studies were conducted in Europe (n = 21) (8, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34), followed by Eastern Asia (n = 14) (35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48) and North America (n = 6) (49, 50, 51, 52, 53, 54). All studies assessed patients of all age groups, with the mean age of onset of vHL being 30.6 years (as reported in 13 publications (17, 24, 30, 31, 39, 41, 43, 46, 49, 50, 52, 55, 58)) and the mean onset age of PPGL being 28.7 years (8, 21, 24, 26, 31, 35, 40, 44, 47, 49, 50, 52, 55, 58). Nineteen studies sampled their participants consecutively (8, 16, 17, 18, 19, 24, 27, 29, 31, 32, 37, 38, 39, 43, 46, 49, 50, 51, 52), whereas in 11 studies a convenience sample was assessed (20, 21, 22, 25, 34, 35, 45, 47, 53, 56, 57). A total of 1340 participants – from 25 studies – had information regarding their vHL family history status, of whom 1059 reported having family history of vHL disease (15, 17, 22, 23, 24, 26, 28, 29, 30, 34, 37, 40, 41, 42, 43, 44, 45, 46, 47, 48, 52, 53, 56, 57, 58). Description of studies can be consulted in Supplementary Table 3, and information regarding other phenotypic manifestations is displayed in Supplementary Table 4.

The meta-analytical frequency of PPGL among participants with vHL was 19.4% (95% CI = 15.9–23.6%), with severe heterogeneity being detected (I2 = 86.1%, Q Cochran P  value ≤ 0.001) (Table 1). In meta-regression analysis (Table 2), studies published in more recent years were associated with a significantly increased frequency of PPGL (OR = 1.04, 95% CI = 1.01–1.08, P = 0.013). No statistically significant differences were observed regarding the mean age of assessed patients (OR = 1.05, 95% CI = 0.96–1.16, P = 0.300). When assessing the frequency of PPGL in the general cohort of vHL participants, we found significant heterogeneity in all subgroup analyses. Similar results were observed in sensitivity analyses restricted to studies with more than 25 participants (Supplementary Table 5). Similar results were also observed when pooling the meta-analytical frequency of pheochromocytoma (excluding PPGLs) among vHL participants assessed in studies with adequate definition of this outcome (Supplementary Table 6).

Table 1

Meta-analytical results on the frequency of pheochromocytoma and/or paraganglioma (PPGL) among patients with von Hippel–Lindau disease (vHL).

	N studies	N vHL	Frequency (%, 95% CI)	Q Cochran P value	I²
Participants with vHL (all vHL types)
Frequency of participants with PPGL	45	4263	19.4 (15.9–23.6)	<0.001	86.1%
Participants with bilateral pheochromocytoma	14		59.5 (47.0–70.9)	0.611	0%
Males with PPGL	15		58.3 (47.6–68.4)	0.837	0%
Participants with vHL type 2		840
Frequency of participants with PPGL	52		60.0 (53.4–66.3)	<0.001	54.6%
Participants with PPGL in vHL type 2A	28	411	58.2 (49.7–66.3)	0.010	36.2%
Participants with PPGL in vHL type 2B	24	256	49.8 (39.9–59.7)	0.013	42.7%
Participants with PPGL in vHL type 2C	11	52	84.1 (75.1–93.1)	0.932	0%

Table 2

Results of meta-regression and subgroup analysis concerning the frequency of pheochromocytoma and/or paraganglioma (PPGL) among patients with von Hippel–Lindau disease (vHL) of all types.

	Meta-regression				Subgroup analysis
	N studies	N vHL	OR (95% CI)	P value	Frequency (%, 95% CI)	Q Cochran P value	I²
Publication year	45	4263	1.04 (1.01–1.08)	0.013*	–a	–a	–a
Sample size	45	4263	1.00 (1.00–1.00)	0.818	–a	–a	–a
% of males	24	648	1.01 (0.98–1.03)	0.554	–a	–a	–a
Mean cohort age	9	444	1.05 (0.96–1.16)	0.300	–a	–a	–a
Mean onset age of vHL	13	1863	0.98 (0.91–1.07)	0.697	–a	–a	–a
Region	45	4263	–	0.802b
Europe	21	2148	–c	–c	19.1 (14.5–24.9)	<0.001	84.4%
Asia and Pacific	16	1116	0.97 (0.55–1.71)	0.923	18.5 (12.9–25.9)	<0.001	76.7%
America	8	999	1.23 (0.62–2.44)	0.558	21.0 (11.4–35.3)	<0.001	92.6%
Sampling method	45	4263	–	0.352b
Consecutive	19	3684	–c	–c	20.5 (16.3–25.5)	<0.001	90.1%
Convenience	11	178	1.22 (0.61–2.41)	0.576	22.4 (11.7–38.7)	<0.001	69.4%
Not specified	15	401	0.72 (0.40–1.28)	0.261	15.3 (9.9–23.0)	<0.001	61.2%
Diagnosis method	45	4263	–	0.888b
Included genetic testing/not specified	42	4086	–c	–c	19.7 (16.1– 23.8)	<0.001	85.0%
Clinical criteria only	3	177	0.93 (0.32–2.70)	0.888	27.7 (3.2–81.6)	<0.001	91.8%

aNot performed, as this is a continuous variable; bOmnibus P  value for this variable; cReference category; *Statistically significant P  value.

The pooled frequency of asymptomatic carriers of VHL pathogenic mutations was 6.2% (95% CI = 3.7–10.3%, I2 = 84.7%, P ≤ 0.001) (15, 16, 17, 21, 23, 24, 25, 26, 27, 28, 30, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58). Among participants with pheochromocytoma, 59.5% had bilateral synchronous or metachronous disease (95% CI = 47.0–70.9%, I2 = 0%, P = 0.611) (20, 22, 25, 26, 28, 30, 35, 40, 41, 45, 47, 55, 57, 58). In addition, 58.3% participants with PPGL were males (95% CI = 47.6–68.4%, I2 = 0%, P = 0.837) (20, 25, 26, 32, 35, 37, 41, 43, 44, 45, 47, 52, 53, 56, 58).

Frequency of PPGL among participants with vHL type 2

A total of 52 studies, assessing 840 participants with vHL type 2, provided information on the frequency of PPGL. Most of these studies were conducted in Europe (n =18) (16, 17, 21, 25, 26, 28, 29, 62, 66, 68, 70, 71, 72, 73, 76, 83, 91, 92), followed by Eastern Asia (n =17) (35, 36, 40, 44, 45, 47, 48, 74, 75, 77, 78, 79, 80, 81, 82, 86, 87) and North America (n =9) (53, 59, 60, 61, 63, 64, 65, 69, 84). All studies assessed patients of all age groups, with the mean onset age of PPGL being 28.8 years (as reported in 21 studies (21, 26, 35, 40, 44, 47, 55, 59, 60, 61, 65, 67, 71, 72, 74, 77, 79, 85, 86, 87, 88)). Six studies reported a consecutive sample (16, 17, 29, 66, 77, 78), whereas in 39 studies a convenience sample was assessed (21, 25, 35, 45, 47, 53, 56, 59, 60, 61, 62, 63, 64, 65, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93).

The frequency of PPGL among participants with vHL type 2 was 60.0% (95% CI = 53.4–66.3%), with moderate heterogeneity being observed (I2 = 54.6%) (16, 17, 21, 25, 26, 28, 29, 35, 36, 40, 44, 45, 47, 48, 53, 55, 56, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93) (Table 1). We observed lower heterogeneity in subgroup analyses restricted to convenience sampling (I2 = 25.6%) and to patients whose diagnosis was performed only clinically (I2 = 23.1%) (Supplementary Table 7). Similar results were obtained with sensitivity analyses restricted to studies with more than 25 participants (Supplementary Table 5), as well as when assessing the meta-analytical frequency of pheochromocytoma without considering PPGLs (Supplementary Table 6).

Twenty-eight studies, with a total of 411 participants with vHL type 2A, reported the frequency of PPGL among these patients (16, 25, 26, 28, 29, 35, 36, 47, 55, 60, 61, 62, 63, 65, 69, 70, 71, 72, 73, 74, 75, 78, 79, 80, 81, 82, 85, 92). The pooled frequency of PPGL among participants with vHL type 2A was 58.2% (95% CI = 49.7–66.3%, I2 = 36.2%) (Table 1) (16, 25, 26, 28, 29, 35, 36, 47, 55, 60, 61, 62, 63, 65, 69, 70, 71, 72, 73, 74, 75, 78, 79, 80, 81, 82, 85, 92). In most subgroup analysis, moderate heterogeneity was observed (Table 3).

Table 3

Results of meta-regression and subgroup analysis concerning the frequency of pheochromocytoma and/or paraganglioma (PPGL) among patients with von Hippel–Lindau disease (vHL) type 2A.

	Meta-regression				Subgroup analysis
	N studies	N vHL	OR (95% CI)	P value	Frequency (%, 95% CI)	Q Cochran p value	I²
Publication year	28	411	0.98 (0.93–1.03)	0.384	–a	–a	–a
Sample size	28	411	1.00 (0.99–1.01)	0.573	–a	–a	–a
Region	28	411	–	0.306b	–a	–a	–a
Europe	11	255	–c	–c	59.6 (48.9–69.5)	<0.165	35.4%
Asia and Pacific	11	69	0.60 (0.25–1.41)	0.239	48.8 (32.3–65.5)	0.135	32.2%
America	6	87	1.29 (0.48–3.44)	0.617	73.4 (37.2–92.8)	0.009	80.1%
Sampling method	28	411	–	0.118b
Consecutive	3	24	–c	–c	55.8 (24.3–83.2)	0.142	50.3%
Convenience	21	370	0.99 (0.32–3.01)	0.981	56.1 (47.4–64.5)	0.012	33.9%
Not specified	4	17	5.39 (0.81–35.76)	0.081	87.6 (61.0– 97.0)	0.767	0%
Diagnosis	28	411	–	0.427b
Included genetic testing/not specified	27	409	–c	–c	57.9 (49.3–66.0)	0.008	36.9%
Clinical criteria only	1	2	2.69 (0.38–19.08)	0.427	83.3 (19.4–99.0)	<0.001	0%

aNot performed, as this is a continuous variable; bOmnibus P  value for this variable; cReference category.

Twenty-four studies, with a total of 256 participants with vHL type 2B, reported the frequency of PPGL among these patients (21, 29, 36, 40, 44, 45, 48, 53, 55, 56, 64, 66, 67, 68, 72, 77, 78, 83, 84, 86, 88, 89, 91, 93). The meta-analytical frequency of PPGL among participants with vHL type 2B was 49.8% (95% CI = 39.9–59.7%, I2 = 42.7%) (21, 29, 36, 40, 44, 45, 48, 53, 55, 56, 64, 66, 67, 68, 72, 77, 78, 83, 84, 86, 88, 89, 91, 93) (Table 1). Low or undetectable heterogeneity was observed when performing subgroup analyses restricted to European studies (pooled frequency = 37.8%, 95% CI = 27.8–48.9%, P = 0.778, I2 = 0%) or to studies using convenience sampling (pooled frequency = 55.9%, 95% CI = 46.1–65.2%, P = 0.135, I2 = 12.7%) (Table 4). All other tested covariates were associated with moderate heterogeneity.

Table 4

Results of meta-regression and subgroup analysis concerning the frequency of pheochromocytoma and/or paraganglioma (PPGL) among patients with von Hippel–Lindau disease (vHL) type 2B.

	Meta-regression				Subgroup analysis
	N studies	N vHL	OR (95% CI)	P value	Frequency (%, 95% CI)	Cochran P value	I²
Publication year	24	256	1.00 (0.95–1.05)	0.965	–a	–a	–a
Sample size	24	256	1.00 (0.96–1.04)	0.975	–a	–a	–a
Region	24	256	–	0.138b	–a	–a	–a
Europe	7	83	–c	–c	37.8 (27.8–48.9)	0.778	0%
Asia and Pacific	11	105	1.97 (0.85–4.57)	0.112	55.0 (39.3–70.0)	0.066	45.2%
America	6	68	2.55 (0.91–7.12)	0.074	58.8 (30.4–82.4)	0.052	58.0%
Sampling method	24	256	–	0.361b
Consecutive	4	62	–c	–c	45.6 (24.0–69.0)	0.094	59.1%
Convenience	15	155	1.48 (0.58–3.80)	0.412	55.9 (46.1–65.2)	0.135	12.7%
Not specified	5	39	0.72 (0.21–2.51)	0.606	40.8 (17.2–69.4)	0.118	47.7%
Diagnosis	24	256	–	0.301b
Included genetic testing/not specified	20	222	–c	–c	46.2 (35.8–56.9)	0.021	44.4%
Clinical criteria only	4	34	2.04 (0.70–5.92)	0.301	73.5 (45.7–90.0)	0.272	29.6%

aNot performed, as this is a continuous variable; bOmnibus P  value for this variable; cReference category.

Lastly, in 11 studies, with a total of 52 participants with vHL type 2C, information was provided on the frequency of PPGL among these patients (25, 26, 29, 40, 44, 45, 59, 74, 76, 87, 90). The meta-analytical frequency was 84.1% (95%CI = 75.1–93.1%), with no heterogeneity observed (I2 = 0%) (25, 26, 29, 40, 44, 45, 59, 74, 76, 87, 90) (Table 1).

Risk of bias assessment

All studies had at least one high risk of bias item concerning internal validity (Fig. 2). Ten studies (12.5%) had low risk of bias in all items concerning external validity (16, 17, 18, 24, 27, 29, 31, 38, 49, 51). The two items for which studies were most commonly classified as having 'low risk of bias' concerned the possibility of nonresponse bias (97.5%) and the presentation of an appropriate numerator and denominator for the parameter of interest (97.5%). Conversely, the case definition was the item for which studies were most frequently classified as having 'high risk of bias', with only two studies (2.5%) using an acceptable case definition (55, 71). 44% of all included studies were considered representative of the target population. However, this percentage rises to 73% when analyzing only those 45 cohort studies used in the meta-analysis pooling the frequency of PPGL in patients with vHL of any subtype. On the other hand, only 23% studies are considered representative of the target population when considering solely those studies assessing the frequency of PPGL in patients with vHL type 2.

Figure 2

Risk of bias graph depicting the assessment of the methodological quality of included primary studies. Risk of study bias items assessed: aStudy’s target population as a close representation of the vHL national population. bSampling frame as a true or close representation of the target population. cSample selection based on a random/consecutive process. dLikelihood of nonresponse bias. eAcceptable case definition used in the study. fSame mode of evaluation for all subjects. gNumerator and denominator appropriate.

Discussion

In this systematic review, we assessed the frequency of PPGL in patients with vHL, as well as in patients with each subtype of vHL. The meta-analytical frequency of PPGL in patients with vHL was 19%, which is in agreement with previous estimates (11, 31). This frequency increased to 60% in vHL type 2 participants and was determined to be 58, 50, and 84% among participants with vHL type 2A, type 2B, and type 2C, respectively. Most participants who developed pheochromocytoma had bilateral adrenal involvement (60%), including both synchronous and metachronous processes. Overall, these results point to the high frequency of PPGL among patients with vHL (particularly in patients belonging to type 2 families, in whom this frequency is exceptionally high), as well as to a high propensity for bilateral adrenal involvement, supporting the benefit of lifetime screening for PPGL in patients with vHL. Meta-regression analysis revealed that more recent studies were associated with a higher frequency of PPGL, which is likely related to an improvement in diagnostic procedures for PPGL and to the expanding application of annual screening protocols in vHL patients. We found no statistically significant differences regarding the mean age of assessed patients – in this setting, it would be expected for studies assessing cohorts of older patients to be associated with a higher frequency of PPGL, as the disease penetrance increases with age, being estimated to be as high as 95% by 60 years of age (49). However, the interpretation of this analysis may be limited by the fact that we used the mean cohort age rather than individual patient-level information regarding participants’ age. Additionally, it should be noted that the mean age of onset of PPGL of 28.7 years, in the upper end of the third decade of life, probably reflects a substantial number of clinically diagnosed cases, compared to the genetic screening and surveillance that increasingly allow for earlier diagnosis.

This study has some limitations. Firstly, most primary studies did not present clear outcome definitions, with some using the term 'pheochromocytoma' to broadly refer to tumors arising from adrenal and extra-adrenal chromaffin cells, others using the term to strictly refer to tumors of adrenomedullary origin, and others not appropriately defining the term. To tackle this limitation, our outcome variable corresponded to PPGL rather than pheochromocytoma. Secondly, some primary studies reported single-institution experiences with small sample series, possibly consisting of a reduced number of families, which conveys a risk of overrepresentation of certain vHL subtypes – to account for this limitation, we performed sensitivity analyses restricted to studies with more than 25 participants, with similar results being observed. Additionally, we came across some studies that disclosed only the number of families with vHL type 2 phenotype rather than the number of participants with PPGL – these reports could not be included in the quantitative synthesis. Another limitation concerns the possibility of selection bias, particularly in the calculation of the frequency of PPGL in vHL type 2C participants – as discussed earlier, to account for the fact that PPGL is the only phenotypic manifestation in the rare vHL type 2C and, therefore, the diagnostic feature in the clinical diagnosis of this subtype, we included only primary studies that have tested asymptomatic members of vHL type 2C families for germline pathogenic VHL mutations; however, in most cases, the testing coverage of the pedigree was not disclosed and therefore it is likely that some members were not tested. This fact combined with the limited number of studies assessing vHL type 2C participants may have contributed to an overestimation of the frequency of PPGL in this subtype. Another limitation concerns the recently described germline mutations in the E1’ cryptic exon of the VHL gene in patients previously considered mutation negative, consisting of a small subset of patients possibly not assessed in the studies included in this meta-analysis (94). Another limitation concerns the lack of important information provided by primary studies, namely regarding sociodemographic data of the vHL participants and adequate characterization of PPGL when these occurred (i.e. mean age of onset, data regarding recurrences or second primary tumors, and data regarding biochemical and imaging procedures). As a result, we were not able to assess several variables that could potentially contribute to explain heterogeneity. Another limitation concerns the fact that information regarding the rate of metastatic PPGL arising in vHL patients was not included in the scope of the query used in this study. In result, the majority of included studies did not provide specific information regarding this subject, preventing from drawing reliable conclusions on this topic. Finally, another important limitation concerns the methodological quality of included primary studies, with all of them having at least one high risk of bias item concerning internal validity.

This study also has several strengths. We conducted a search using three electronic bibliographic databases and did not apply date- or language-based exclusion criteria, so as to minimize the risk of publication bias. We also performed meta-regression and subgroup analysis to identify variables responsible for explaining across-studies heterogeneity. In addition, we assessed the methodological quality of included primary studies, pointing that future primary studies should particularly minimize the risk of bias related to the use of equivocal outcome definitions and disclose more detailed patient-level information, namely regarding the development and coexpression of manifestations, for further establishment of phenotypic associations in frame with the genetic findings. Finally, the results of this systematic review have clinical relevance, as they provide additional information for the endocrinological management of vHL patients and their respective families, supporting the benefit of early and lifetime screening for PPGL, as the review points that PPGL is more frequent than that acknowledged in some of the previous reports, particularly among type 2 vHL families.

Conclusions

In conclusion, PPGL is frequent among patients with vHL, with this frequency being higher among patients with vHL type 2 (in whom it ranges, according to the subtype, from 50 to 84%). Overall, this points to a high burden of PPGL among patients with vHL, indirectly supporting early and lifetime screening. Nevertheless, our results should be carefully interpreted on account of the observed heterogeneity and on the important methodological limitations observed in primary studies. For the future, an eventual development of an international vHL register could contribute to overcome such limitations, as it would allow for clinicians and researchers to access genotypic–phenotypic information, benefiting the individual and collective management of the vHL disease, as a day-to-day clinical tool and as a data record for future studies.

Declaration of interest

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.

Funding

This work did not receive any specific grant from any funding agency in the public, commercial, or not-for-profit sector.

Data sharing statement

Additional data will be available for other researchers upon reasonable request to the corresponding author.