Survival in primary hyperparathyroidism over five decades (1965–2010) a population-based retrospective study

Abstract

Context:

Survival in patients with primary hyperparathyroidism (PHPT) remains uncertain.

Objective:

To update survival in patients with PHPT in a United States community population.

Design:

Retrospective cohort study.

Setting:

Community population in Rochester, Minnesota.

Participants:

Residents who met criteria for PHPT from 1965 to 2010.

Interventions:

Survival was estimated using the Kaplan Meier product-limit method. The Cox proportional hazards model was used to determine associations, as relative hazards (RR) with 95% confidence intervals (CI), of various risk factors with time to death.

Main outcome measure:

The overall age and gender-adjusted survival compared to white Minnesota residents.

Results:

We identified 1139 PHPT individuals, 76% female, with a median age of 58 years. Most were observed without parathyroidectomy (69%). The relative risk of death among the entire cohort was 0.996 (95% CI: 0.91–1.09, P = 0.935) which was not different compared to Minnesota residents. Those with maximum serum calcium level ≥ 10.8 mg/dL (0.7 mg/dL above the reference range) had an increase in mortality (RR 1.32, 95% CI: 1.10–1.58, P = 0.002). Survival among all PHPT individuals after parathyroidectomy was no different from expected (RR = 1.06, 95% CI 0.89–1.28; P = 0.508). Mortality was significantly decreased after parathyroidectomy in those with serum calcium levels ≥10.8 mg/dL (HR 0.47, 95% CI: 0.36–0.61, P < 0.001).

Conclusions:

Mortality in the entire cohort was not different from expected. PHPT patients with a maximum serum calcium level ≥ 10.8 mg/dL had increased mortality. Survival was improved after parathyroidectomy in those with this degree of hypercalcemia.

1. Introduction

Primary hyperparathyroidism (PHPT) is a common endocrine disorder which has demonstrated a recent increased incidence rate [1]. In the modern era, most PHPT patients are asymptomatic (89%) and discovered incidentally based on a mildly elevated serum calcium measurement [1]. The current management of these patients without symptoms or classic complications of PHPT has been guided by expert consensus recommendations [2] due to the lack of large randomized controlled trials to inform the impact of surgery on disease complications and overall survival [3]. These guidelines advocate hypercalcemia >1 mg/dL above the reference range as one criteria to be used as a consideration for parathyroidectomy, although evidence supporting clinically significant benefit of surgery on disease complications or survival is generally inconclusive.

The association of PHPT with increased mortality and the beneficial impact of parathyroidectomy on survival are controversial. Evidence from European cohorts has demonstrated a reduction in expected survival in PHPT treated surgically [4–7] and in cohorts observed without surgery [8,9]. Other studies found no decrease in survival after surgery [10] or found an association of surgery with a reduction in mortality risk [11]. Previous data from our community-based PHPT cohort in North America indicated that overall survival was not adversely affected subsequent to a diagnosis of PHPT [12]. The association of biochemical indices of PHPT severity with survival in modern cohorts is also uncertain. PTH levels were correlated with mortality in one study [13] but not another [11]. Similarly, calcium levels as a risk for increased mortality have been reported in some studies [11,12,14] but not all [13,15].

The purpose of the current study was to update survival in unselected patients with PHPT in our cohort and to determine if there has been a significant change in overall PHPT mortality, and if biochemical parameters of PHPT are associated with survival. The large number of observed patients with mild disease in our cohort also permitted assessment of the risk of death in the typical patient identified with PHPT in the current era. Availability of the population-based medical records-linkage system of the Rochester Epidemiology Project offered a unique opportunity to address this clinically important patient outcome [16,17].

2. Methods

After approval from the Institutional Review Boards of Mayo Clinic and the Olmsted Medical Center, this comprehensive medical records-linkage system (the Rochester Epidemiology Project) was used to identify all Olmsted County, Minnesota, residents with PHPT from 2002 through 2010 as previously described [1]. Cases were then followed to allow adequate time since diagnosis to provide more meaningful results of mortality. Recorded diagnoses and surgical procedures are indexed including at autopsy or on death certificates. All Olmsted County residents with serum calcium levels exceeding 10.1 mg/dL (2.52 mmol/L) at least twice between 2002 and 2010 were identified from Mayo’s Laboratory Information System. The assays used to measure serum calcium concentrations have changed since 1965, however the laboratory verifies the applicability of the reference interval with each change. The reference interval for serum calcium did not change over the time period (1965–2010), therefore the serum calcium concentrations measured over the time period should be comparable.

Total calcium values uncorrected for albumin were used. PTH assay (two-site immunochemiluminometric assays) reference ranges during the study period were as follows: 1.0–5.2 pmol/L by manual bead immunoassay (in-house assay) in December 1974 to December 2, 2003; 10–55 pg/mL (1.1 to 5.8 pmol/L) by Nichols Advantage (Nichols Institute Diagnostics, San Juan Capistrano, CA, U.S.A.) in December 3, 2003 to May 1, 2006; 10–67 pg/mL (1.1 to 7.1 pmol/L) by Diagnostics Products Corporation assay (Diagnostics Products Corporation, Los Angeles, CA, U.S.A.) in March 31, 2005 to July 15, 2007; and 15 to 65 pg/mL (1.59 to 6.90 pmol/L) performed on the Roche Cobas 6000 (Roche Diagnostics GmbH, Mannheim, Germany) since July 16, 2007.

For each potential case from the complete community medical record was reviewed by one of the investigators (MLG). Mayo Clinic records contain details of every inpatient hospitalization, outpatient clinic visit, emergency room and nursing home care, and all radiographic and pathology reports, including autopsies [18]. This information was supplemented by that available from other providers for local residents, including Olmsted Medical Center [16], the second largest health care system in the county.

Inclusion criteria for patients for PHPT were the same as in our previous reports [1,19,20]. Patients were accepted as having “definite” PHPT if they met one or more of the following criteria: i) histopathologic proof of parathyroid adenoma or hyperplasia; ii) hypercalcemia (calcium level > 10.1 mg/dL) with an elevated serum PTH; or iii) hypercalcemia persisting for >1 year with no other cause (such as: thiazide diuretics, cancer, creatinine level > 176.8 mmol/L, or lithium therapy, granulomatous disease, etc.) who did not have PTH measurements. Two groups of patients were identified as having “possible” hyperparathyroidism: i) patients with at least two elevated serum calcium levels from at least three determinations who were followed for <1 year; ii) patients with at least two elevated serum calcium levels and in at least two different years that were followed by three or more normal calcium values. Patients carrying a diagnosis of familial benign hypocalciuric hypercalcemia or multiple endocrine neoplasia were excluded. To assure comparability with the earlier studies and minimize any potential impact of ill patients with unrecognized disease moving into the community for medical care, patients must have established residency in Rochester for at least one year before the initial elevated serum calcium level was identified.

The updated cohort from 2002 to 2010 was combined with the previously identified Rochester PHPT cases [19–21]. Additionally, PHPT patients were analyzed in four time periods based on periods of incidence rate change identified within our population: Pre-chemistry panel era, 1965 to June 30, 1974; early chemistry panel era, July 1, 1974 through 1984; late chemistry panel era, 1985 through 1997; and osteoporosis screening era, 1998 through 2010 [1]. Patient characteristics evaluated included sex, age, maximum serum calcium level, and management with parathyroidectomy. Cause-specific deaths were determined on the basis of the diagnosis listed first on the death certificate and were allocated into 4 subsets were according to International Classification of Diseases, Tenth Revision (ICD-10) circulatory system diseases (includes cardiovascular, cerebrovascular, and other vascular causes) I00–I99; all malignant cancers C00–C97; respiratory diseases (includes respiratory infections and chronic respiratory diseases) and all other causes.

Group characteristics for continuous variables were expressed as mean ± standard deviation (SD) or median [interquartile-range] according to distribution characteristics and compared using ANOVA/Student’s t-test/Wilcoxon-test as appropriate. Qualitative data are expressed as percentage and compared using the chi-square/Fisher’s exact test. Survival after diagnosis was estimated using the Kaplan Meier product-limit method. Comparison of survival experience between two or more groups was done with log-rank test. The Cox proportional hazards model was used to determine associations, as relative hazards (RR) with 95% confidence intervals (CI), of various risk factors with time to death. Survival after the diagnosis of PHPT in community residents was compared with expected rates of Minnesota white residents, adjusted for age and gender of the PHPT subjects using the one-sample log-rank test. The alpha level was set at 0.05 for all statistical analyses and analyzed using SAS © (9.4) and R-3.2.3. Cox proportional hazard model was used to compute relative risk of mortality for PHPT cohort in comparison to Rochester Epidemiology Project (REP) cohort. REP resources (database and programing algorithm) were used to construct comparison cohort with simple random 1:1 matching from all residence of the REP on index date, age and gender. Randomly matched cohort from REP database is used as a control group of the local population. For cause specific death, any other type of death was censored.

3. Results

From 1965 through 2010, we identified 1139 PHPT patients in Rochester, Minnesota, of whom the majority had definite PHPT (94%) (Table 1). Spanning over 5 decades, the updated cohort has added 18 years of study, 10,066 person years of follow-up after the initial elevated serum calcium level, and 704 PHPT patients. These patients had typical demographic characteristics of PHPT with a median age of 57.8 years and the majority were female (75.9%). Overall, the hypercalcemia was mild with a median maximum serum calcium level of 10.7 mg/dL and only 21% of subjects had a maximum serum calcium level ≥ 11.2 mg/dL (≥1 mg/dL above our reference range). Most patients did not undergo parathyroidectomy (69.3%). Across the different periods of PHPT incidence rate changes, we observed that patients in the most recent era were older, and compared to the previous time periods, less likely to have possible PHPT (Table 1).

Table 1.

Demographic and clinical characteristics of Rochester, Minnesota, residents with a first diagnosis of primary hyperparathyroidism (PHPT) in 1965–2010, based on periods of incident rate changes.

Characteristics	1965-June 1974	July 1974–1984	1985–1997	1998–2010	Total	P-value*
	(N = 64)	(N = 369)	(N = 194)	(N = 512)	(N = 1139)
Age						<0.001
Mean (SD)	53.1 (15.7)	55.7 (16.3)	55.1 (16.4)	60.4 (14.1)	57.6 (15.5)
Median	55.2	57.3	55.1	60.6	57.8
Q1, Q3	43.1, 65.4	47.6, 67.9	45.0, 66.5	51.4, 70.7	48.5, 68.9
Sex, n (%)						0.286
Female	46 (71.9%)	284 (77.0%)	138 (71.1%)	396 (77.3%)	864 (75.9%)
Male	18 (28.1%)	85 (23.0%)	56 (28.9%)	116 (22.7%)	275 (24.1%)
Ethnicity/race, n (%)						<0.0001
White	29 (45.3%)	216 (58.5%)	156 (80.4%)	440 (85.9%)	841 (73.8%)
Other/unknown	35 (54.7%)	153 (41.5%)	38 (19.6%)	72 (14.1%)	298 (26.2%)
Mode of diagnosis, n (%)						<0.001
Histology	22 (34.4%)	95 (25.7%)	50 (25.9%)	182 (35.9%)	349 (30.8%)
Inappropriate PTH level	27 (42.2%)	161 (43.6%)	79 (40.9%)	251 (49.5%)	518 (45.7%)
Hypercalcemia >1 yr	13 (20.3%)	87 (23.6%)	48 (24.9%)	52 (10.3%)	200 (17.7%)
Possible PHPT	2 (3.1%)	26 (7.0%)	16 (8.3%)	22 (4.3%)	66 (5.8%)
Maximum serum calcium level (mg/dL)						0.002
Median	10.8	10.7	10.7	10.8	10.7
25th-75th percentiles	10.6, 11.3	10.5, 11.1	10.5, 11.0	10.6, 11.1	10.5, 11.1
Range	(10.2–12.5)	(10.2–16.0)	(10.2–13.1)	(9.4–14.7)+	(9.4–16.0)
Serum calcium ≥ 11.2 mg/dL, n (%)	19 (29.7%)	78 (21.1%)	32 (16.5%)	109 (21.3%)	238 (20.9%)	0.149
Surgery	23 (35.9%)	97 (26.3%)	50 (25.8%)	180 (35.2%)	350 (30.7%)	0.011

^*P-values comparing characteristics of PHPT patients diagnosed in 1998–2010 to those diagnosed in 1985–1997.

⁺Includes 2 patients with normocalcemic PHPT with confirmed parathyroid pathology.

Patients were followed for 15,746 person-years after the initial elevated serum calcium level (1614 person-years for 1965–June 1974; 7805 person-years for July 1974–1984; 2644 person-years for 1985–1997; 3683 person-years for 1998–2010). A total of 483 patients (42.4%) have died upon last follow up. The overall age and gender-adjusted survival in patients with PHPT, relative risk of death of 0.996 (95% CI: 0.91–1.09, P = 0.935), was no different compared to expected for white Minnesota residents (Fig. 1). Sensitivity analysis with exclusion of possible PHPT individuals yielded similar results, relative risk of death 1.03 (95% CI: 0.94–1.13, P = 0.497) This implies that the results of the primary analysis were not affected by the presence of patients with possible PHPT. Predictors of mortality in a univariate analysis were older age at time of the initial serum calcium elevation, male sex, maximum serum calcium elevation, and diagnosis of PHPT after June 1974 (Table 2). In a multivariate model older age, male sex, maximum serum calcium level, and PHPT diagnosis after June 1974 (RR 1.48, 95% CI: 1.30–1.68, P < 0.001) remained significantly associated with all-cause mortality. Minnesota population cause specific mortality rates are not available. Therefore, we explored cause specific mortality using the smaller population of the Rochester Epidemiology Project (REP). Age and gender adjusted relative risk of death by 4 broad categories of disease was based on the PHPT patients matched 1:1 to a REP sample control cohort. The relative risk of death among PHPT was increased for circulatory system-related (RR 1.58 (95% CI:1.25–2.0) P = 0.002) and respiratory-related (RR 1.37 (95% CI:1.37–2.61) P = 0.0001) but not for cancer-related (RR 1.19 (95% CI: 0.89–1.6) P = 0.236) or other cause of death (RR 1.14 (95% CI:0.925–1.41) P = 0.215). All-cause mortality compared to matched REP control group was increased (RR 1.37 (95% CI: 1.20–1.56) P ≤0.0001).

Fig. 1.

Overall survival among Rochester, Minnesota residents with primary hyperparathyroidism from 1965 through 2010 compared with that expected for Minnesota white residents of similar age and gender (RR 0.996, 95% CI: 0.911–1.09, P = 0.935).

Table 2.

Predictors of all-cause mortality among Rochester, Minnesota residents with primary hyperparathyroidism diagnosed from 1965 through 2010.

Univariate model
Variable	HR	95% CI of HR		P-value
Age (per 5 year increase)	1.62	1.55	1.70	<0.001
Male sex	0.75	0.60	0.94	0.014
Maximum serum calcium	1.46	1.26	1.69	<0.001
Time of detection (period):
Reference = 1964-June 1974
July 1974–1984	1.96	1.39	2.76	<0.001
1985–1997	3.36	2.20	5.12	<0.001
1998–2010	7.97	5.15	12.32	<0.001
Multivariable model
Variable	HR	95% CI of HR		P-value
Age (per 5 year increase)	1.58	1.51	1.66	<0.001
Male sex	1.28	1.02	1.61	0.032
Maximum serum calcium	1.21	1.02	1.42	0.028
Time of detection (period):
Reference = 1964-June 1974
July 1974–1984	1.32	0.94	1.85	0.112
1984–1997	1.86	1.23	2.82	0.004
1998–2010	3.07	2.00	4.71	<0.001

Since maximum serum calcium was an independent predictor of mortality, this was explored further by plotting survival based on quartiles of serum calcium (Fig. 2). That is to say, the cohort was divided into 4 groups with approximately equal number of individuals (quartiles) based on their maximum serum calcium level. Therefore, the range of serum calcium values is not equal in the groups. Survival progressively declined with increasing quartiles of serum calcium elevation, with the worst survival among patients with maximum serum calcium levels above 10.7 mg/dL. Patients with serum calcium levels ≥10.8 mg/dL had a significant increase in mortality versus those with levels below 10.8 mg/dL (RR 1.32, 95% CI: 1.10–1.58, P = 0.002). Cut point analysis of the relationship of risk of death and calcium level for the entire cohort showed that the risk of mortality exceeded 1 at a calcium of 10.8 mg/dL and continued to increase at higher calcium levels (Fig. 3A). Similar analysis restricted to those with calcium values <12 mg/dL showed an expected change to the slope but a crossing of the relative risk of death of 1.0 appears at a similar level of calcium as in the overall group (Fig. 3B). When comparing patients with serum calcium levels below 10.8 mg/dL to those with levels 10.8 mg/dL or higher, patients with higher levels included more females, were older and more likely to undergo parathyroidectomy, and had more deaths (Table 3). PTH levels ≤ the upper limit of normal compared to ≤ twice the upper limit of normal and > twice the upper limit of normal were not significantly associated with mortality (P = 0.889). To further evaluate the increased mortality in those diagnosed after June 1974, mortality was analyzed in the periods of incident rate changes and no significant decrease was found in expected survival during these specific intervals (1965–June 1974, RR 0.93, 95% CI: 0.70–1.23, P = 0.590; July 1974–1984, RR 0.98, 95% CI: 0.86–1.10, P = 0.700; 1985–1997, RR 1.00, 95% CI: 0.79–1.26, P = 1.000; 1998–2010, RR 1.09, 95% CI: 0.90–1.31, P = 0.399).

Fig. 2.

Survival based on quartile of serum calcium (10.2–10.5 mg/dL; 10.6–10.7 mg/dL; 10.8–11.1 mg/dL; >11.1 mg/dL) among Rochester, Minnesota residents with primary hyperparathyroidism from 1965 through 2010.

Fig. 3.

Relationship of risk of death to calcium level. Spline plots displaying unadjusted hazard ratios of all-cause mortality with serum calcium values as a continuous variable. A. Entire cohort; B. cohort restricted to those with calcium <12 mg/dL. Note the scale of the axes are different between A and B. Dotted lines indicate 95% CIs.

Table 3.

Demographic and clinical characteristics of Rochester, Minnesota, residents with a first diagnosis of primary hyperparathyroidism in 1965–2010, by severity of hypercalcemia.

Characteristic	Serum calcium < 10.8 mg/dL	Serum calcium ≥ 10.8 mg/dL	Overall	P-value*
	(n = 588)	(n = 551)	(n = 1139)
AGE [median (IQR)]	55.3 (46.7–66.7)	60.1 (51.4–70.9)	57.8 (48.5–68.9)	<0.001
Male [n (%)]	161 (27.4%)	114 (20.7%)	275 (24.1%)	0.008
Parathyroid surgery [n (%)]	89 (15.1%)	261 (47.4%)	250 (30.7%)	<0.001
Death [n (%)]	221 (37.6%)	262 (47.5%)	483 (42.4%)	<0.001
Years to last follow-up/death from detection [median (IQR)]	12 (6.3–20.7)	10.9 (6.2–18.2)	11.5 (6.3–19.8)	0.083

^*<10.8 mg/dL compared to ≥10.8 mg/dL.

Parathyroid surgery was performed in 350 patients (30.7%), with more patients undergoing surgery from 1965 to June 1974 (35.9%) and 1998 to 2010 (35.2%) compared to those patients from July 1974 to 1997 (26.1%). The cumulative parathyroid surgery events after the initial serum calcium elevation were highest in the most recent era followed by 1964 to June 1974. However, cumulative parathyroidectomies were only significantly different in the most recent era compared to the middle two periods (P < 0.001). The mean (±SD) time between the initial serum calcium elevation and parathyroidectomy was 3.09 ± 4.05 years. Patients undergoing parathyroid surgery had significantly higher maximum serum calcium elevations (median 11 mg/dL) compared to those who were observed (10.7 mg/dL) (P < 0.001) and were also significantly younger (mean age 55.9 years versus 58.3 years; P = 0.003). Patients were followed for 4668 person-years after parathyroidectomy (92 person-years for 1965–June 1974; 815 person-years for July 1974–1984; 1494 person-years for 1985–1997; 2267 person-years for 1998–2010). Survival among all PHPT patients after parathyroidectomy was no different from expected (RR = 1.06, 95% CI 0.89–1.28; P = 0.508) (Fig. 4). However, among those undergoing parathyroidectomy, mortality was significantly decreased after parathyroidectomy in patients with serum calcium levels ≥10.8 mg/dL (HR 0.47, 95% CI: 0.36–0.61, P < 0.001) compared to those with calcium levels <10.8 mg/dL. Mortality was not significantly different after surgery in those with serum calcium levels <10.8 mg/dL (HR 0.89, 95% CI: 0.58–1.36, P = 0.582).

Fig. 4.

Survival among Rochester Minnesota residents with primary hyperparathyroidism after parathyroid surgery compared with that expected for Minnesota white residents of similar age and gender (RR 1.06, 95% CI: 0.89–1.28, P = 0.508).

4. Discussion

We report that survival was not decreased after hypercalcemia was detected in this updated U.S. community-based cohort of PHPT patients. Although mortality in the overall cohort was the same as expected, we identified a relationship between the magnitude of hypercalcemia, a biochemical marker of disease severity, and mortality. Maximum calcium greater than or equal to 10.8 mg/dL (0.7 mg/dL above the reference range) was associated with an increased risk of death. Indeed, each increasing quartile of maximum serum calcium was associated with reduced survival. In addition, mortality in those with maximum serum calcium levels 10.8 mg/dL–11.1 mg/dL closely paralleled the mortality observed in those with levels ≥11.2 mg/dL, suggesting that lower levels of hypercalcemia may be clinically relevant in regard to an increased risk of death. Previously, we found in the earlier eras of the cohort from 1965 to 1992 that only individuals in the highest quartile of calcium (≥11.2 mg/dL) had a significantly worse survival than the lower three quartiles [12]. Then as now, most of the patients had mild PHPT that was identified on routine chemistry panel screening.

In contrast to the survival data on PHPT patients in the US, European and Australian cohorts largely based on patients with more severe PHPT undergoing parathyroidectomy have demonstrated an increased risk of death. [4–7,15]. However, European PHPT cohorts comprised of observed patients with mild to moderate disease also demonstrated a reduction in expected survival [8,9]. More recently, a retrospective population-based study in Tayside, Scotland from 1997 to 2006 of observed patients with mild PHPT, which is more comparable to our Rochester cohort, demonstrated an increased risk of all-cause mortality (SMR 2.62; 95% CI 2.39–2.86) [22]. The same cohort was also matched to five population-based age, gender, and calendar year of PHPT diagnosis controls, and once again, an increased risk of mortality was observed [23].

The relationship between degree of hypercalcemia and mortality in PHPT has not been consistently found in other cohorts. An Australian cohort identified from 1961 to 1994 [15] reported that serum calcium levels above 12 mg/L (3 mmol/L) were not associated with reduced survival. In the Tayside population-based cohort, calcium was only associated with mortality in the short-term whereas baseline PTH was associated both in the short- and long-term [13]. We did not find a relationship between PTH level, another biochemical marker of disease severity, and death. In contrast, data from a Danish case-control study of patients undergoing parathyroidectomy with more severe hypercalcemia showed that higher serum calcium levels in those operated on from 1979 to 1990 had reduced survival when compared to those patients undergoing parathyroidectomy with lesser calcium elevations from 1991 to 1997 [14]. A single institution cohort from Scotland was recently studied and a higher adjusted calcium was associated with mortality in the entire cohort and in the patients not treated with surgery [11].

We explored cause specific mortality in our PHPT cohort compared to the Rochester Epidemiology Project population. In contrast to our earlier report, we found an increased risk of mortality in the categories of vascular diseases (which includes cardiovascular and cerebrovascular disease as well as other vascular diseases) and respiratory diseases but not cancer or other causes. While the specific cause of death associated with mildly elevated serum calcium levels in PHPT is not well understood, observational studies have suggested that serum calcium levels higher within the reference range in the population are positively linked to an increased risk of cardiovascular disease [24] and that PTH and calcium are independently associated with subclinical vascular disease [25]. In addition, several studies have identified an increased risk of cardiovascular death in PHPT [15,22,23,26] and a couple studies found no association between serum calcium levels and risk for death [13,15]. A recent randomized controlled trial of parathyroidectomy in mild PHPT, reported a decrease in some cardiovascular risk factors after parathyroidectomy. However, only patients with a higher calcium had a significant reduction in pulse wave velocity, a measure of arterial stiffness, after surgery compared to the control group [27]. While our findings of an increase in vascular mortality are broadly congruent with the Tayside findings, the exact groupings of cause of death are not identical [22]. In addition, the Tayside analysis excluded more severe PHPT whereas our cohort represents the entire spectrum of disease severity in our population.

We also examined the impact of parathyroidectomy on survival and did not find a significant effect of parathyroidectomy on mortality in the overall PHPT cohort. However, we did observe a significant reduction in mortality in patients undergoing parathyroidectomy with serum calcium levels ≥ 10.8 mg/dL. This has important implications since the most recent guidelines on management of asymptomatic PHPT suggest that serum calcium level > 1 mg/dL (0.25 mM) the upper limit of normal is an indication to consider parathyroid surgery [2]. A recently reported study of a cohort of inpatients in Scotland with PHPT found increased mortality and a reduction in mortality in those treated with parathyroid surgery [28]. Biochemical parameters of disease severity were not included in this Scottish cohort analysis. Lastly, we did not collect information regarding complications of PHPT such as diagnosis of osteoporosis or nephrolithiasis which may have impacted decision making in our cohort regarding pursuing parathyroidectomy.

Unfortunately, there is a lack of evidence from randomized trials for the management recommendation of parathyroidectomy compared to observation in regard to some important outcomes, including mortality [3]. While one study reported that the increased mortality risk normalized approximately 5 years after parathyroidectomy in patients with milder PHPT operated on in 1980 [29], other reports have suggested a persistently elevated risk of death several years after parathyroidectomy [4,30]. Others did not observe any significant change in mortality after parathyroidectomy [15]. Survival was not decreased in PHPT patients referred to our institution from 1980 to 1984 who underwent parathyroidectomy, as compared to the expected survival in the Midwest population [10].

It remains unclear why mortality in our updated cohort increased since the earliest time period. Compared to pre-July 1974 PHPT patients, each subsequent era had worsened survival, which was significant from 1984 to 1997 and 1998 to 2010, despite the fact that pre-1974 patients were more likely to present with symptomatic disease [1]. Others have also demonstrated that the period of diagnosis influences survival such that an earlier era of PHPT has better survival compared to a more recent diagnosis [15]. Of note, the Tayside cohort (1997–2006) [22] coincided with the most recent era of our cohort (1998–2010), and although this era had greater mortality compared to earlier time periods, this subgroup was older and had less follow-up. Despite the concern about reduced survival in PHPT with more recent cohorts, mortality was still not different from expected in our population during the most recent time period (RR 1.085, 95% CI: 0.897–1.313, P = 0.399). We cannot exclude that shifts in medical care for other conditions over the decades in our cohort may impact survival in PHPT.

Our study has several inherent limitations including the retrospective design. As such, we did not have routine assessment of PHPT related complications such as osteoporosis or other factors that might be related to mortality risk. We used “uncorrected” total serum calcium measurements which have superior accuracy to albumin adjusted calcium [31] although ionized calcium may be more sensitive to diagnosing hypercalcemia in mild PHPT with normal total calcium [32]. In addition, the population of Olmsted County is primarily white, which limits the application of study results to more ethnically diverse populations [33]. The recording of ethnicity/race in the medical record was not universal, but more common in the recent cohort. This is notable since PHPT appears to be more common among blacks compared to other races [34]. The similarity of Olmsted County population characteristics to those of Minnesota and the upper Midwest has been shown in prior data. And, mortality rate among Olmsted County residents is similar to the United States overall [35]. Although some recent data may suggest that Olmsted County may be slightly healthier than state residents overall [36]. Finally, it is not possible to dissect whether the increase in mortality in the most recent years of our cohort is due to confounding by shifts in overall medical care in the 50 years of study.

In summary, in a well-defined unselected U.S. community-based cohort of PHPT subjects with predominantly asymptomatic disease and mild hypercalcemia mortality is not increased. Unlike previously, we now find an increase in vascular disease related mortality compared to Rochester Epidemiology Project population. However, risk factors for death in PHPT include the degree of hypercalcemia, such that higher maximum serum calcium levels are associated with reduced survival. Based on the increased mortality associated with lower levels of hypercalcemia identified in this cohort combined with the observed reduction in mortality in those PHPT patients with a maximum serum calcium ≥ 10.8 mg/dL (0.7 mg/dL above the reference range), the current recommended threshold for the degree of hypercalcemia to recommend parathyroidectomy may be too high [2]. Randomized prospective studies of parathyroidectomy versus observation would more clearly determine surgical benefits for important patient outcomes such as mortality. Further research is needed to better understand the differences reported between U.S. PHPT patient mortality compared to those from outside the U.S., especially given the recent increase incidence of this disease.

Abbreviations:

HR

hazard ratio

PHPT

primary hyperparathyroidism

PTH

parathyroid hormone

RR

relative risk

SD

standard deviation

95% CI

95% confidence interval

Data availability

The datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.