Introduction
Pheochromocytomas and paragangliomas (PPGLs) are rare neuroendocrine tumors arising from the adrenal medulla or extra-adrenal paraganglia, respectively, which produce catecholamines [
1]. PPGLs can present in many ways, and symptoms and signs can be difficult to interpret initially [
2]. With the increasing use of imaging, more PPGLs are found as incidentalomas, i.e., masses found on imaging studies ordered for unrelated conditions [
2]. In patients with adrenal incidentalomas, 0.6–4.2% are found to have a PPGL [
3‐
5]. PPGLs are potentially fatal if not diagnosed and/or managed appropriately [
6,
7].
Once diagnosed, the pre-operative treatment of choice is an alpha-blocker with progressive dose up-titration for at least 1–2 weeks before surgery to prevent peri-operative cardiovascular complications [
6]. Laparoscopic surgery is usually recommended for smaller (<6 cm) pheochromocytomas and suitable paragangliomas, while open surgery is recommended for larger tumors, or if laparoscopic surgery fails [
6]. Hypertension, paroxysmal or consistent, is very common and metabolic disturbances, such as diabetes, are fairly prevalent [
1,
8]. In a large proportion of patients, hypertension has been reported to persist after surgery [
9,
10], while metabolic derangements were more likely to improve [
11]. Occasionally, both diabetes and hypertension resolve after surgery, although it is unclear how often. The histopathological diagnosis of PPGLs is typically straightforward, but discrimination between benign and malignant lesions is extremely difficult. The presence of metastasis is the only certain way to diagnose a malignant PPGL [
12]. The prognosis of paragangliomas is considered worse compared to pheochromocytomas, and life-long follow-up has been suggested [
1,
6].
The majority of PPGLs have previously been found due to symptoms of elevated catecholamines, but with increasing use of high-resolution imaging techniques, incidentalomas are becoming more common. We have recently reported that the majority of pheochromocytomas nowadays are found during the investigation of incidentalomas [
2]. Moreover, family screening for genetic syndromes has increased, including those related to PPGLs (e.g., multiple endocrine neoplasia type 2 [MEN2A], Von Hippel Lindau syndrome [VHL], neurofibromatosis type 1 [NF1], and mutations in succinate dehydrogenase B, C, and D [SDHx]). Thus, PPGL treatment and outcomes could be expected to have changed over time and may not be as previous literature suggests.
The aims of the present study were to investigate in a large cohort of PPGLs the management, short-term (surgical complications, hypertension, and diabetes), and long-term outcomes (recurrence, metastasis, and mortality), and to determine if these are different depending on the mode of presentation, or tumor type.
Materials and methods
Eligible for inclusion were all patients with an International Classification of Diseases version 10 (ICD-10) code of E27.5 (adrenomedullary hyperfunction) and/or C74.1 (malignant neoplasm of medulla of adrenal gland) being admitted and/or attending the outpatient clinic between June 2005 and January 2018 at the Department of Endocrinology, Metabolism and Diabetes, Karolinska University Hospital, Stockholm, Sweden. The patients’ electronic medical files were reviewed and if a PPGL could not be confirmed the case was excluded. The diagnosis was confirmed using both imaging and urine/plasma catecholamines, in addition to the histological result post surgery. At the time of the review, The National Population Register was also consulted to retrieve the date of death if applicable [
13]. Attending physicians in Sweden code all hospital admissions and specialist outpatient visits with ICD-10 codes, and these are stored in both local and national databases [
14]. The mode of presentation (incidentaloma, symptomatic PPGL [defined as a patient suspected to have PPGL due to symptoms and signs before biochemical confirmation and imaging], screening, mixed adrenaline and noradrenaline secretion, only noradrenaline secretion), final diagnosis (pheochromocytoma or paraganglioma), tumor size, peri-operative management, length of stay (from the day of surgery to discharge), histology (KI67 and suspicion of malignancy), blood pressure, glucose abnormalities (both at the time of PPGL diagnosis and at the first endocrine review 6–12 months after surgery), genetic results if found (many results were not automatically incorporated into the medical electronic files), follow-up time, mortality including cause of death, metastasis from the PPGL, and recurrence were noted. The screening group consisted of patients with a known familial disease that could cause PPGLs, and therefore underwent regular screening for PPGL. Catecholamines were measured using high-performance liquid chromatography (HPLC) for 24 h urinary adrenaline and noradrenaline (normal <80 and <400 nmol/24 h, respectively), and liquid chromatography–tandem mass spectrometry (LC/MS/MS) for plasma metanephrine and normetanephrine (normal <0.3 and <0.6 nmol/L, respectively). As not all individuals had had both tests performed, the highest urine or plasma level was divided with the upper level of normal and noted. Only noradrenaline secretion was defined as adrenaline and/or metanephrine levels being below the upper limit of normal. The definition suggested by Esienhofer et al. [
15], i.e., noradrenergic tumors being defined by a tumor-derived increase in plasma normetanephrine with either a lack of increase in plasma metanephrine or an increase of metanephrine less than 5% that of both normetanephrine and metanephrine, could not be used since we did not have enough data on plasma metanephrines. The recorded blood pressures used in this study were measured with an appropriate-sized blood pressure cuff on two occasions at rest, mostly seated in an office or occasionally on the ward. KI67 was determined using standardized methodology in the clinically accredited pathology laboratory by staining with an anti-KI67 antibody and counting manually 2000 cells in hotspots. Evaluation of histology using the Pheochromocytoma of the Adrenal Gland Scaled Score (PASS) [
12] or Grading System for Adrenal Pheochromocytoma and Paraganglioma (GAPP) was done in some cases, but not all. The definition of suspicion of malignancy was a PASS score ≥4, a GAPP score ≥3, or if none of these score systems had been used, if the pathologist wrote suspicion of malignancy and from the report it seemed likely that the score was ≥4 or ≥3, respectively. Improvement of blood pressure post surgery was defined as a reduction of systolic and diastolic blood pressure of at least 10 mmHg, together with and/or reduction in blood pressure medications. Prediabetes was defined as HbA1c 42–47 mmol/mol and/or fasting plasma glucose 6–6.9 mmol/L and/or random plasma glucose 7.8–11 mmol/L. Part of this cohort has been used in a previous study of the initial clinical presentation of pheochromocytomas [
2].
The Regional Ethical Review Board in Stockholm, Sweden, approved the study and as this was a retrospective study, formal consent was not required.
Statistical analysis
Mean ± SD (if normally distributed) or median and interquartile range (25–75%) were used. Two groups and continuous variables were compared with unpaired t-test (normally distributed) or Mann–Whitney rank-sum test, and three groups with one-way ANOVA (normal distributions) or with ANOVA on ranks test. In frequency table calculations, Chi-square or Fisher’s exact test were used, whichever was appropriate. Correlations were calculated using linear regression analysis. A P value <0.05 was considered significant. SigmaStat 3.0 for Windows (Systat Software Inc., San Jose, California) was used for all calculations.
Discussion
This large study reinforces some results of previous studies but also demonstrates novel findings, including that mode of presentation may predict outcomes. Blood pressure and glycemic abnormalities improved in most patients, but recurrence, PPGL-related metastasis, and PPGL-related deaths still occurred, confirming the necessity for long-term follow-up.
Retrospective studies have shown that alpha-blockers are the preferred choice to reduce peri-operative complications [
6]. Phenoxybenzamine is a long-acting, non-selective alpha-blocker which has traditionally been used, but short-acting selective alpha-blockers, such as prazosin, terazosin, and doxazosin, are becoming more popular due to fewer side-effects [
6]. The vast majority of our patients were on doxazosin which was up-titrated to doses sometimes higher than the recommended final dose of 32 mg/day [
6]. The dose or time on alpha-blockers did not differ between how the patient presented or between the pheochromocytomas and paragangliomas. Laparoscopic approach was the standard approach in all groups except in the paraganglioma group. However, while almost half of the paraganglioma group had an initial laparoscopic approach, two-thirds of them required conversion to an open procedure and the length of hospital stay post surgery was twice as long. Laparoscopic surgery for paragangliomas has been claimed as safe with outcomes similar to laparoscopic surgery for pheochromocytomas [
19]. In contrast, our results indicate that a laparoscopic approach should be used only in select patients, which is in accordance with international guidelines [
6]. Otherwise, aside from a tendency to longer post-operative stays in the symptomatic PPGL group, no differences were found between the different groups concerning peri-operative outcomes.
Hypertension, both sustained and paroxysmal, is common in PPGLs, with up to 90% affected at diagnosis [
1,
2,
20]. Hypertension in PPGL has been considered potentially curable with surgery. We found a dramatic drop in blood pressure with surgery for all PPGLs, with the majority being able to cease all blood pressure medications and having normalized blood pressure postoperatively. The drop was similar in all subgroups, except for those found by screening, probably due to normal blood pressure at diagnosis. However, the paraganglioma group received the same amount of antihypertensive medication post surgery as at diagnosis, despite normalization of blood pressure. Our improvement in blood pressure seems much better than others who have shown persistent hypertension in 51–79% at last follow-up [
9,
10]. Weismann et al. [
10] reported that only 38% had improved blood pressure using similar criteria as us, compared to 94% of our patients. On the other hand, our blood pressure follow-up was done 6–12 months post surgery, and we did not have long-term follow-up data of blood pressure. Our patients were probably not as symptomatic, since less than a third presented with suspicion of PPGL, compared to 57 to >85% reported by the other research groups [
9,
10,
20]. Nevertheless, all our patients in the symptomatic PPGL group also had blood pressure improvement by surgery. It has been reported that of 46 patients with hypertension after PPGL surgery at long-term follow-up, 30 (65%) were already hypertensive at 1 year post surgery. In our 6–12 months follow-up data, the majority were normotensive, suggesting that PPGL surgery may cure hypertension in most patients nowadays.
Glycemic disturbances are common at PPGL diagnosis [
1,
11,
20,
21]. We found 26% of our patients with PPGL were affected by diabetes, which is slightly less than the 40% given in reviews [
1], but similar to the 31% found in the most recent included patients by Amar et al. [
20], and in a recent Japanese study [
21]. Interestingly, in the last study, all patients underwent an oral glucose tolerance test (OGTT) and 4 of 13 (31%) included patients had impaired glucose tolerance, compared to our 21%. In contrast, we did not perform regular OGTTs but diagnosed prediabetes on fasting, random glucose levels, and/or HbA1c, which likely explains this difference. Of note, our symptomatic PPGL group had a higher frequency of diabetes compared to the incidentaloma or screening groups, but the latter two groups had more prediabetes, such that overall glycemic disturbances were similar. It could be speculated that the shift in presentation of PPGL will ameliorate the glycemic disturbances at presentation even more in the future. Like other studies, we found a dramatic improvement in insulin sensitivity by surgical treatment [
11,
21], and most of our patients had no signs of diabetes post surgery, especially those on diet control preoperatively. Those with persistent diabetes could mostly reduce their insulin doses by >50%, or decrease the number of oral antidiabetic drugs, while still maintaining improved glycemic control. Thus, surgery in PPGL can lead to resolution of diabetes, or at least dramatically improve glycemic control. Interestingly, in our study the prevalence of glycemic disturbance was almost four times more common in the pheochromocytoma group compared to the paraganglioma group, and the latter group did not improve much by surgical treatment. Such a comparison has never been made previously [
8]. Adrenaline has been reported to impair glucose metabolism more than noradrenaline [
8]. Since functional paragangliomas almost exclusively produce noradrenaline, this could explain the difference.
The histological discrimination between benign and malignant tumors is challenging, with metastasis being the definition of a malignant PPGL according to WHO [
12]. By this definition, 5% of our cases were malignant, with almost twice that number in the PPGL suspicion group. Around a forth of our cases were suspected malignant. The fact that not all cases were assessed formally using the PASS or GAPP scores could be viewed as a limitation, however, scoring systems have not been found to be reproducible and reliable, and their utility in predicting future metastasis has not been proven [
12]. KI67 is a marker being used in many tumors for predicting malignancies. Most of our patients had very low KI67, but the symptomatic PPGL group had a median value of 2.5%. One study reported that two out of four cases with a KI67 >2% had future metastasis, compared to three out of 41 with lower KI67 [
22], which is similar to our findings. Thus, although KI67 index may be useful as a risk estimator at group level, it is not valuable to predict metastatic development in individual patients given the large number of false-positive and false-negative results. In contrast, tumor size seemed more valuable in predicting future metastasis with a cutoff value of 9.5 cm in our study. In guidelines, the importance of size for the prediction of metastasis have been emphasized [
6,
23].
The long-term outcomes appear reasonable in our study, with 13% having either recurrence or metastasis. This is slightly lower than others have reported [
20], although when they divided their PPGL cohort into quartiles based on date of operation, patients in the most recent quartile (surgery around 2000) had similar figures to ours. In a previous similar study, metastasis, but not cardiovascular disease, was reported to reduce life expectancy [
24]. Of the 64 patients with presumed benign PPGL, seven (11%) developed PPGL-related metastasis during follow-up, and all died. In our cohort, only two of the five developing PPGL-related metastasis died, five of the total cohort died of other malignancies, and three cardiovascular deaths occurred, of which two probably were at least partly PPGL-related. However, in almost a third of the deaths in our study, we did not identify a cause of death, which is similar to the previously mentioned study [
24]. Thus, the long-term mortality in PPGL seems to predominately be due to PPGL-related metastasis and other malignancies, but longer follow-up with better documentation of the cause of death is needed to evaluate this further.
The only significant difference in long-term outcomes between the different subgroups in our study was an increased recurrence rate in the screening group, which is in accordance with others [
20]. However, looking at the absolute numbers, the symptomatic PPGL group seemed to have a slightly worse prognosis, but the numbers were small. Moreover, the cause of death in the symptomatic PPGL group was 50% PPGL-related, while none in the incidentaloma or screening group had a PPGL-related death. This pattern has not previously been investigated. In contrast to what is normally stated in review articles and guidelines [
1,
6], we could not demonstrate any worse prognosis in paragangliomas compared to pheochromocytomas. This was encouraging, but it should be noted that we only included paragangliomas with catecholamine excess due to the ICD-10 codes used, i.e., nonfunctional paragangliomas were excluded since they usually receive other ICD-10 codes. Moreover, the paraganglioma group was quite small so the study may have been underpowered to show a difference.
Almost all our paragangliomas secreted only noradrenaline but most pheochromocytomas had mixed adrenaline and noradrenaline secretion, which is similar to others [
24,
25]. The outcomes between mixed and only noradrenaline secreting tumors were not different, however, those with mixed secretion were older and had higher rates of diabetes at presentation.
Like all retrospective studies, there are several limitations, in particular that of ascertainment bias. Not all cases had had a genetic evaluation. While this is a large study compared with similar single center studies, some subgroups were quite small, thus results from these subgroups must be interpreted with caution. Furthermore, we were not able to standardize the measurements and follow-up due to the retrospective nature of the study, however, the follow-up of mortality was complete thanks to the comprehensive coverage of The National Population Register.
In conclusion, the mode of presentation may be associated with short- and long-term outcomes. Our outcomes overall seem slightly better than previous studies. Short-time outcomes were slightly better for pheochromocytomas compared to paragangliomas, but long-term outcomes were similar. All PPGLs benefit from treatment, but earlier diagnosis may be better. Long-term follow-up is necessary [
23].