MANAGEMENT OF ENDOCRINE DISEASE: Recurrence or new tumors after complete resection of pheochromocytomas and paragangliomas: a systematic review and meta-analysis

in European Journal of Endocrinology
Correspondence should be addressed to O Steichen; Email: olivier.steichen@aphp.fr
*(L Amar and C Lussey-Lepoutre contributed equally to this work)

Objectives

To systematically review the incidence and factors associated with recurrences or new tumors after apparent complete resection of pheochromocytoma or thoraco–abdomino–pelvic paraganglioma.

Design

A systematic review and meta-analysis of published literature was performed.

Methods

Pubmed and Embase from 1980 to 2012 were searched for studies published in English on patients with non-metastatic pheochromocytoma or thoraco–abdomino–pelvic paraganglioma, complete tumor resection, postoperative follow-up exceeding 1 month, and recurrence or new tumor documented by pathology, hormonal dosages, or imaging tests. Incidence rates of new events after curative surgery were calculated for each study that had sufficient information and pooled using random-effect meta-analysis.

Results

In total, 38 studies were selected from 3518 references, of which 36 reported retrospective cohorts from the USA, Europe, and Asia. Patient follow-up was neither standardized nor exhaustive in the included studies. A clear description of patient retrieval methods was available for nine studies and the follow-up protocol and patient flow for four studies. Only two studies used multivariable methods to assess potential predictors of postoperative events.

The overall rate of recurrent disease from 34 studies was 0.98 events/100 person-years (95% confidence interval 0.71, 1.25). Syndromic diseases and paragangliomas were consistently associated with a higher risk of a new event in individual studies and in meta-regression analysis.

Conclusions

The risk of recurrent disease after complete resection of pheochromocytoma may be lower than that previously estimated, corresponding to five events for 100 patients followed up for 5 years after complete resection. Risk stratification is required to tailor the follow-up protocol after complete resection of a pheochromocytoma or paraganglioma. Large multicenter studies are needed to this end.

Abstract

Objectives

To systematically review the incidence and factors associated with recurrences or new tumors after apparent complete resection of pheochromocytoma or thoraco–abdomino–pelvic paraganglioma.

Design

A systematic review and meta-analysis of published literature was performed.

Methods

Pubmed and Embase from 1980 to 2012 were searched for studies published in English on patients with non-metastatic pheochromocytoma or thoraco–abdomino–pelvic paraganglioma, complete tumor resection, postoperative follow-up exceeding 1 month, and recurrence or new tumor documented by pathology, hormonal dosages, or imaging tests. Incidence rates of new events after curative surgery were calculated for each study that had sufficient information and pooled using random-effect meta-analysis.

Results

In total, 38 studies were selected from 3518 references, of which 36 reported retrospective cohorts from the USA, Europe, and Asia. Patient follow-up was neither standardized nor exhaustive in the included studies. A clear description of patient retrieval methods was available for nine studies and the follow-up protocol and patient flow for four studies. Only two studies used multivariable methods to assess potential predictors of postoperative events.

The overall rate of recurrent disease from 34 studies was 0.98 events/100 person-years (95% confidence interval 0.71, 1.25). Syndromic diseases and paragangliomas were consistently associated with a higher risk of a new event in individual studies and in meta-regression analysis.

Conclusions

The risk of recurrent disease after complete resection of pheochromocytoma may be lower than that previously estimated, corresponding to five events for 100 patients followed up for 5 years after complete resection. Risk stratification is required to tailor the follow-up protocol after complete resection of a pheochromocytoma or paraganglioma. Large multicenter studies are needed to this end.

Introduction

Pheochromocytomas and paragangliomas (PH/PG) are rare neuroendocrine tumors arising from chromaffin cells of the adrenal medulla or sympathetic and parasympathetic paraganglia respectively. Following resection of the primary tumor, most patients are free of tumor with no clinical, biochemical, or imaging evidence of persistent disease. However, tumor-free patients are at a risk of long-term recurrence, defined as the reappearance of disease after complete surgical eradication of the tumor (1). Recurrences may arise at the operated site or may be metastatic, developing in non-chromaffin organs, mainly the lymph nodes, bones, lungs, and liver (2, 3). Patients with inherited tumors may also develop new PH/PGs in the contralateral adrenal gland or in other paraganglia (3, 4). At least 15% of patients undergoing surgery for PH/PG develop new tumors or recurrences, most of which are metastatic (4). Consequently, long-term follow-up is recommended for patients who have undergone surgery for PH/PG (5).

Although there are reports of the prognostic value of various clinical, genetic, and pathological features, there are no robust prognostic indices of recurrence other than the higher probability of new events in patients with inherited tumors and possibly in patients with extra-adrenal or large tumors (4, 5). The total duration of follow-up that is required remains unclear, as new events may occur decades after initial surgery. The optimal combination and sequence of biochemical and imaging tests to detect and monitor recurrences are poorly defined. There have been no comprehensive systematic reviews assessing the incidence of recurrences or new tumors following surgery for PH/PG.

Our primary objective is to systematically review the incidence of local or metastatic recurrences or new tumors in patients who have undergone apparently complete resection of a non-metastatic PH/PG. A secondary objective is to assess the factors associated with recurrences (local or metastatic) or new tumors.

Methods

Eligibility criteria

We searched randomized or non-randomized controlled trials, prospective or retrospective cohort studies, and case–control studies published in English in 1980 or later (computed tomography and metanephrine determinations were not universally available before 1980). Studies were eligible if (i) they enrolled at least 20 patients with PH/PG; (ii) patients had reportedly undergone complete tumor resection; (iii) postoperative follow-up exceeded 1month (arbitrarily); and (iv) the number of patients with recurrence or new tumor could be identified.

Information sources and search

We searched Medline and Embase databases from 01/01/1980 to 10/19/2012. We developed a specific search strategy for each database and is presented in Supplementary Methods (see section on supplementary data given at the end of this article).

Study selection

Two physicians, experts in endocrine hypertension (L A and C L), independently made a first selection using the titles, abstracts, and keywords to exclude studies that clearly did not fulfill the inclusion criteria. They independently read the full text of the remaining papers to identify eligible articles. At each stage, disagreements between readers were resolved by discussion or, if necessary, by a third reader (P F P or J L). Journals and authors were not blinded during study selection. In cases of overlapping publications by a given team, only the most comprehensive or most recent was included.

The studies were then further subdivided into two categories: (i) studies on head and neck paraganglioma only and (ii) studies on pheochromocytoma and thoraco–abdomino–pelvic paraganglioma. This review reports only the results of the latter category of studies.

Data collection

A standardized form was tested on ten articles and two senior experts (P F P and J L) to homogenize further data collection. The form was then used by two abstractors (L A and C L) to extract methodological and clinical data from all included studies. Records were reviewed by a third reader and issues were resolved by discussion. Journals, titles, and authors were not blinded during data abstraction.

The following data were collected from each study: first author and publication year; study design and settings; number of centers, total number and number of operated patients with complete tumor resection; risk of bias; percentage of female patients, mean age, percentage of hypertensive patients, percentage of patients with pheochromocytoma, location of paragangliomas, mean tumor size, percentage of secreting tumors, and percentage of genetic or syndromic diseases; number of patients with follow-up and duration of follow-up; incidence of recurrence, timing of recurrence, and attributable deaths.

Quality assessment

No agreed criteria exist for assessing the risk of bias of prognostic studies. We used several resources to compile a list of criteria to assess the risk of bias regarding study participants, prognostic factors, outcome, follow-up, and reporting. Further details are provided in the Supplementary Methods.

Summary measures

Incidence rates of recurrence or new tumor over the entire duration of the follow-up were calculated by dividing the number of events by the number of person-years of follow-up, and then standardized to the number of events/100 person-years. The mean duration of follow-up was multiplied by the number of patients when the number of person-years of follow-up was not reported. When the mean duration of follow-up was not reported, it was approximated by a formula using the median and range when available (6, 7), or by the median in other cases. We assumed the Poisson distribution to estimate 95% confidence intervals for the rate of events presented in forest plots. Meta-analysis was performed on rates and their confidence interval limits. Heterogeneity was assessed using Cochran statistic and the I² inconsistency coefficient. An I² value greater than 50% was considered to be indicative of substantial heterogeneity. Study results were pooled using random-effect meta-analysis. We used Stata SE/MP version 9.2 (StataCorp) for the analyses.

Additional analyses

We performed sensitivity analyses based on the availability of the mean duration of follow-up (allowing to compute the exact incidence rate) and by risk of bias. A funnel plot of the event rate according to the number of person-years of follow-up (a substitute for the precision of estimates) was used to assess publication bias.

Meta-regression analysis was used to assess the impact of the following variables on the logarithm of incidence rates: operation year of the first patient, specialty of the author team (surgery vs others), percentage of females, mean age, percentage of familial diseases (genetic or syndromic), percentage of pheochromocytomas, mean size of the tumor, and mean duration of follow-up.

Results

Study selection

The details of the selection process and reasons for exclusions are depicted in the flow chart (Fig. 1). Bibliographic searches yielded 3518 references, among which 42 on pheochromocytoma and thoraco–abdomino–pelvic paraganglioma were finally selected, reporting 38 different cohorts.

Figure 1
Figure 1

Study flow diagram. FU, follow-up; HN, head and neck; PH/PG, pheochromocytomas and paragangliomas; PG, paragangliomas.

Citation: European Journal of Endocrinology 175, 4; 10.1530/EJE-16-0189

Download Figure

Study characteristics

The characteristics of the included studies are reported in Table 1.

Table 1

Overview of included studies.

StudyCountryPeriodDesignSettingsNumber of centersTotal number of patientsNumber of curative surgeries
Agarwal 2012 (8)India1990–2010RetrospectiveSurgery1101100
Amar 2005/2006 (9, 10)France1975–2006RetrospectiveMedicine1261242
Beatty 1996 (11)Ireland1970–1991RetrospectiveMixed14138
Benhammou 2010 (12)USA1995–2003RetrospectiveSurgery12626
Brunt 2001/2002 (13, 14)USA1993–2000RetrospectiveSurgery13535
Castilho 2009 (15)Brazil1995–2006RetrospectiveSurgery12424
Cotesta 2009 (16)Italy1992–2008RetrospectiveMedicine19191
de Wailly 2012 (17)France1993–2009RetrospectiveMedicine15348
Diner 2005 (18)USA1995–2004RetrospectiveSurgery13333
Edström Elder 2003 (19, 20)Sweden1976–1999RetrospectiveMixed18580
Favia 1998 (21)Italy1977–1996RetrospectiveSurgery15550
Geoghegan 1998 (22)UK1978–1992RetrospectiveSurgery14342
Grozinsky-Glasberg 2010 (23)Israel1989–2009RetrospectiveMedicine24341
van der Harst 2002 (24)The Netherlands1983–2001RetrospectiveMixed18787
Hayry 2009 (25)Finland1985–2008RetrospectiveLaboratory14236
Iacobone 2011 (26)Italy1985–2010RetrospectiveMedicine17170
Jaroszewski 2003 (27)USA1992–2001RetrospectiveSurgery14747
Kercher 2002 (28)USA1995–2000RetrospectiveSurgery13939
Khorram-Manesh 2005 (29)Sweden1950–1997RetrospectiveMedicine1121118
Lairmore 1993 (30)USA1956–1990RetrospectiveMixed35855
Zhang 2007/Lang 2008 (31, 32)China1998–2005RetrospectiveSurgery1103103
Lucon 1997 (33)Brazil1974–1994RetrospectiveMixed15041
Lumachi 1998 (34)Italy1977–1996RetrospectiveSurgery15550
Neumann 1999 (35)Germany1985–1995RetrospectiveMixed13939
Noshiro 2000 (36)Japan1957–1995RetrospectiveMedicine19591
Obara 1995 (37)Japan1981–1994RetrospectiveSurgery18783
Pan 2005 (38)China1990–2004RetrospectiveSurgery12625
Park 2011 (39)Korea1989–2008RetrospectiveSurgery1152147
Pomares 1998 (40)Spain1979–1995RetrospectiveMedicine14443
Rodriguez 2008 (41)SpainRetrospectiveSurgery25454
Scholten 2011 (42)The Netherlands1959–2010RetrospectiveSurgery16161
Scott 1984 (43)USA1950–1983RetrospectiveSurgery36953
Stenström 1988 (44)Suede1956–1982ProspectiveMixed16464
Tiberio 2008 (45)Italy2000–2006RCTSurgery12222
Timmers 2008 (46)The Netherlands1966–2000RetrospectiveMedicine16964
Tormey 2002 (47)Ireland1989–2000RetrospectiveMedicine213932
Wilhelm 2006 (48)USA1995–2005RetrospectiveSurgery36564
Yip 2004 (49)USA1962–2003RetrospectiveSurgery15958

All studies were on retrospective cohorts except one on a prospective cohort (44) and one that was a randomized controlled trial (45). Of the 38 studies, 20 were performed in Europe, nine in North America, six in Asia, two in South America, and one in the Middle East. The inception year ranged from 1950 to 2000 (median 1985). Studies involved 1–21 centers; 20 studies were performed by surgical teams, ten by medical teams, seven by both, and one by laboratory physicians.

Risk of bias within studies

Detailed risks of bias of the included studies are reported in the Supplementary Table 1. Only nine studies explained how they ensured the completeness of patient retrieval (for instance, through a prospectively maintained research database or administrative databases with diagnostic coding). Only four studies provide a clear and specific description of patient flow (inclusion, operation, cure, and follow-up). Fourteen studies looked for prognostic factors, but only two performed multivariate analyses.

Patient and disease characteristics

Patient and disease characteristics are reported in Table 2. The percentage of females was reported in 35 studies and was between 36 and 67% (median 54%). The mean or median age was reported in 35 studies and was between 27 and 53 years (median 42 years). The percentage of hypertensive patients was reported in 25 studies and was between 3 and 98% (median 66%). The percentage of syndromic or genetic disease was reported in 32 studies. Of the 32 studies, four involved only patients with a syndromic disease (VHL, NF1, or MEN2). In the remaining 28 studies with phenotypic and/or genetic information, a syndromic or familial disease and/or a mutation in a gene predisposing to PH/PG was between 2 and 97% (median 20%).

Table 2

Overview of patient and disease characteristics.

StudyFemales (%)Age (years)HTN (%)Pheo (%)PHPG size (mm)Secreting tumors (%)Genetic diseasesa (%)
Agarwal 2012 (8)43Mean 36 (s.d. 14.6)8182Mean 71 (s.d. 25)
Amar 2005/2006 (9, 10)52Mean 42.5 (s.d. 15)8987Mean 54 (s.d. 27)21
Beatty 1996 (11)54908424
Benhammou 2010 (12)Mean 27100Mean 25100
Brunt 2001/2002 (13, 14)51Mean 42 (s.d. 17)54100Mean 34 (s.d. 13)57
Castilho 2009 (15)42Median 46.5 (range 10–75)83100Median 37 (range 5–120)13
Cotesta 2009 (16)52Mean 48 (range 8–77)92Mean 43 (range 10–110)23
de Wailly 2012 (17)55Mean 53 (range 13–88)100Mean 46 (s.d. 13)96b15
Diner 2005 (18)Mean 38 (range 10–79)100Mean 23 (range 8–50)94
Edström Elder 2003/2003 (19, 20)56Range 14–776182Range 10–14018
Favia 1998 (21)49Median 41 (range 10–63)5891Mean 58 (s.d. 30)7
Geoghegan 1998 (22)67Mean 42 (range 16–73)56100Mean 49 (range 20–130)28
Grozinsky-Glasberg 2010 (23)49Mean 52.6 (range 16–77)4788Mean 49 (s.d. 20)
van der Harst 2002 (24)55Mean 46 (range 9–78)6689Mean 54 (s.d. 31)31
Häyry 2009 (25)Mean 46.595Mean 47 (s.d. 26)86c
Iacobone 2011 (26)52Mean 44.8 (range 15–80)93Mean 54 (s.d. 30)24
Jaroszewski 2003 (27)51Mean 53.1 (range 16–81)100Mean 43 (range 10–85)13
Kercher 2002 (28)64Mean 43 (range 19–59)9597Mean 52 (range 20–121)95b10
Khorram-Manesh 2005 (29)56Mean 47.2 (s.d. 16.8)8493Mean 49 (s.d. 24)88b25
Lairmore 1993 (30)57Mean 32.8 (s.d. 12.2)57100Mean 43 (s.d. 27)100
Zhang 2007/Lang 2008 (31, 32)49Mean 35.8 (s.d. 13.3)100Mean 47 (s.d. 23)
Lucon 1997 (33)56Median 33 (range 10–64)8684Median 70 (range 30–200)97b12
Lumachi 1998 (34)49Mean 41 (range 10–63)45907
Neumann 1999 (35)59Mean 40 (range 10–76)100Mean 40 (range 10–90)67
Noshiro 2000 (36)55Mean 40 (s.d. 14)8815
Obara 1995 (37)56Median 40 (range 11–67)84Mean 62 (s.d. 26)9214
Pan 2005 (38)58Mean 39.5 (s.d. 8.9)85Mean 73 (s.d. 35)0
Park 2011 (39)47Mean 46.5 (range 18–76)6290Mean 67 (s.d. 35)2
Pomares 1998 (40)54Mean 43 (s.d. 13.7)3798Mean 49 (s.d. 26)84b52
Rodriguez 2008 (41)57Mean 37.4 (range 14–71)24100Mean 45 (range 10–120)89100
Scholten 2011 (42)39Mean 33 (s.d. 12.7) 310067100
Scott 1984 (43)58Range 9–79987712
Stenström 1988 (44)53Mean 45 (range 15–79)479420
Tiberio 2008 (45)36Mean 51 (range 34–74)100Mean 40 (range 22–60)
Timmers 2008 (46)64Mean 46.1 (s.d. 15.6)788820
Tormey 2002 (47)38Median 36 (range 8–76)898759
Wilhelm 2006 (48)62Mean 48.5 (s.d. 16.1)68100Mean 40 (s.d. 15)14
Yip 2004 (49)58Median 369597

The percentage of pheochromocytoma was reported in 37 studies and was between 77 and 100% (median 94%). Of these, 14 studies involved only patients with pheochromocytomas, and the percentage of patients with pheochromocytoma was between 77 and 98% (median 89%) in the 23 remaining studies with information on tumor location. The mean or median tumor size was reported in 28 studies and was between 23 and 73 mm (median 48 mm).

Results of individual studies

Raw results of individual studies are reported in Table 3. The number of patients followed up after complete surgical resection was between 22 and 242 (median 52). The mean or median duration of follow-up was available for 34 studies and was between 14 and 180 months (median 84 months). The percentage of recurrent disease over the entire follow-up was between 1 and 34% (median 6%). Eleven studies (51 events) provided individual data on the time from surgery to recurrent disease, with an overall median time to event of 60 months (range 3–204), and two studies (26 events) provided summary results with median times to event of 17 and 29 months (range 5–195). Two studies (7 events) did not provide any information on the time to the recurrent event.

Table 3

Overview of treatment and outcomes.

StudyCured and followed upFollow-up duration (months)Same-site recurrencesOther-site recurrencesMetastasesAttributable death
Agarwal 2012 (8)100Mean 44 (range 3–160)0010
Amar 2005/2006 (9, 10)242Mean 102 (range 45.6–158.4)180180
Beatty 1996 (11)38Mean 842044
Benhammou 2010 (12)26Mean 1113300
Brunt 2001/2002 (13, 14)34Mean 46 (range 2–85)0300
Castilho 2009 (15)24Mean 74 (range 18–150)0000
Cotesta 2009 (16)43Range 6–1922012
de Wailly 2012 (17)48Mean 860110
Diner 2005 (18)33Mean 36 (range 3–108)2000
Edström Elder 2003 (19, 20)80Median 1442032
Favia 1998 (21)50Mean 88 (range 6–232)1100
Geoghegan 1998 (22)41Mean 31 (range 9–120)0001
Grozinsky-Glasberg 2010 (23)410000
van der Harst 2002 (24)87Median 120 (range 3–192)011410
Hayry 2009 (25)36Mean 103 (range 20–284)1030
Iacobone 2011 (26)70Median 126 (range 6–300)201
Jaroszewski 2003 (27)45Mean 41 (range 10–89)0100
Kercher 2002 (28)39Mean 14 (range 1–40)0000
Khorram-Manesh 2005 (29)121Mean 1802164
Lairmore 1993 (30)55Mean 112.8 (range 8.4–342)01200
Zhang 2007/Lang 2008 (31, 32)103Range 5–360000
Lucon 1997 (33)35Mean 33 (range 0.33–192)0000
Lumachi 1998 (34)50Mean 88.2 (range 6–232)1000
Neumann 1999 (35)33Mean 73 (range 16–179)170
Noshiro 2000 (36)74Mean 1171254
Obara 1995 (37)83Median 58 (range 1–164)0401
Pan 2005 (38)25Median 66 (range 24–132)0000
Park 2011 (39)147Mean 41.5 (range 0.9–298)001212
Pomares 1998 (40)42Mean 96 (range 24–216)0010
Rodriguez 2008 (41)54Mean 92.5 (range 12–178)0500
Scholten 2011 (42)61Mean 160.8 (range 1.2–501.6)3180
Scott 1984 (43)53Mean 103 (range 12–348)0054
Stenström 1988 (44)64Mean 139.2 (range 12–324)0220
Tiberio 2008 (45)22Mean 35 (range 18–84)0000
Timmers 2008 (46)64Mean 132 (range 12–456)2077
Tormey 2002 (47)322302
Wilhelm 2006 (48)46Mean 24 (range 1–84)1000
Yip 2004 (49)58Median 717310

Candidate prognostic markers assessed by one or more individual studies are reported in Supplementary Table 2. Paragangliomas (hazard ratios 11 and 8.8 compared with pheochromocytomas) and syndromic diseases (hazard ratios 3.4 and 14 compared with sporadic diseases) were independently associated with an increased risk of a new event after curative surgery in two studies (9, 37) and larger tumor size in one study (hazard ratio 1.2 for each supplementary centimeter in tumor diameter) (9). Paragangliomas were also associated with new events by univariate analyses in one additional study but not in two others. Larger tumor size was also associated with an increased risk of a new event by univariate analysis in an additional study but not in three others.

Synthesis of results

The new event rate estimates of 34 studies with follow-up duration data showed significant heterogeneity (I² 51%, P<0.001; Fig. 2). Random-effect meta-analysis produced an overall new event rate estimate of 0.95 events/100 person-years (95% confidence interval (CI) 0.68, 1.21). Assuming a steady incidence over time, this converts into a 5-year cumulative incidence of 4.7% (95% CI: 3.4, 6.1; Fig. 2), distrib­uted as follows: new tumors 22%, local recurrences 23%, and metastatic recurrences 55%.

Figure 2
Figure 2

Forest plot of new event rates.

Citation: European Journal of Endocrinology 175, 4; 10.1530/EJE-16-0189

Download Figure

The pooled estimate of event rate was as follows:

0.98 events/100 person-years (95% CI: 0.68, 1.29) across studies allowing the calculation of the exact number of person-years of follow-up;

1.18 events/100 person-years (95% CI: 0.71, 1.66) across studies with the lowest risk of bias (at most five risks regarding participants, outcome, follow-up, or results);

1.11 events/100 person-years (95% CI: 0.56, 1.66) across studies fulfilling both conditions.

Risk of bias across studies

The funnel plot of incidence rate against person-years of follow-up shows that the estimated incidence rate of several studies deviates from the overall estimate (Supplementary Fig. 1).

Meta-regression analysis

The percentage of genetic or syndromic disease was associated with the rate of new events (Fig. 3 and Supplementary Fig. 2). The pooled event rate estimate was 0.79 events/100 person-years (95% CI: 0.53, 1.04) for studies with less than 60% genetic or syndromic diseases (I² 40%, P=0.03) and 2.24 events/100 person-years (95% CI: 1.62, 2.87) for studies with 60% or higher genetic or syndromic diseases (I² 0%, P=0.51). A younger mean age was also associated with a higher rate of new events; however, this association disappeared in a multivariate model including both the mean age and the percentage of genetic or syndromic disease. Other variables were not associated with the new event rate in meta-regression analyses (Supplementary Table 3).

Figure 3
Figure 3

Meta-regression analysis of event rates according to the proportion of genetic or syndromic diseases.

Citation: European Journal of Endocrinology 175, 4; 10.1530/EJE-16-0189

Download Figure

Discussion

Summary

Our review suggests that the risk of recurrent disease following complete resection of a pheochromocytoma or a thoraco–abdomino–pelvic paraganglioma is lower than that previously estimated. However, the risk remains non-negligible, approximately 5% per 5years of follow-up, and late events are possible, up to 15years after surgery in the included studies. Paragangliomas and familial disease are the two main independent risk factors of recurrent disease identified by several studies and by meta-regression analyses across studies. The association between the size of the primary tumor and the risk of recurrence after complete removal was weaker.

Limitations of evidence

The reporting of methodological details and clinical results of the included studies was very heterogeneous. Most studies were retrospective, and patient follow-up was neither standardized nor exhaustive. The time pattern of recurrence risk (early vs late) was not assessed, and data were scarce after the first 10 years of follow-up, precluding any firm conclusion beyond this horizon, although recurrent disease is clearly still possible. Risk factors were not consistently assessed across studies and only a few had available genetic data. It is now estimated that around 40% of all PH/PG arise in patients carrying a germline mutation in one of the 13 susceptibility genes identified so far (50). Individual patient data were not available, and meta-regression analyses were limited by an incomplete description of patients and disease in individual studies.

Implications

The overall low frequency of recurrent disease after complete resection of a PH/PG calls into question the need for lifelong follow-up after surgery for all patients. This contradicts the estimates from a previous review of published studies that did not take the varying mean duration of follow-up into account (3). Nonetheless, recurrent disease remains possible even after a long and uneventful follow-up and appears to be more frequent in patients with paragangliomas or familial disease than patients with sporadic pheochromocytoma. The follow-up protocol may need to be tailored according to these and other patient and disease characteristics. In particular, several studies that did not meet our inclusion criteria consistently show that germline mutations, even in apparently sporadic cases, are also associated with more aggressive disease and would benefit from intensive follow-up (5).

The results of this systematic review do not allow any conclusion concerning the necessary length and frequency of follow-up after surgery and further research is warranted. Given the rarity of the disease and the number of candidate risk factors, multicenter studies with consistent documentation of the phenotype and genotype of included patients are needed to overcome the limitations of available evidence and reach the power necessary for multivariate prognostic analyses.

Supplementary data

This is linked to the online version of the paper at http://dx.doi.org/10.1530/EJE-16-0189.

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 research was partly funded by the European Society of Endocrinology.

References

 

Official journal of

European Society of Endocrinology

Sections

Figures

  • Study flow diagram. FU, follow-up; HN, head and neck; PH/PG, pheochromocytomas and paragangliomas; PG, paragangliomas.

    View in gallery
  • Forest plot of new event rates.

    View in gallery
  • Meta-regression analysis of event rates according to the proportion of genetic or syndromic diseases.

    View in gallery

Information

Cited By

PubMed

Google Scholar

Related Articles

Altmetrics

Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 154 154 81
PDF Downloads 36 36 26