Discussion and Conclusions• Imprecision was acceptable for all methods with total CVs of 10.9% or less (Table 1). The E170 always had the lowest total CV and the LIAISON had the highest.• All methods correlated well to the E170 (Figure 1, Table 2). The IMMULITE exhibited the highest positive bias for both sample types, with the largest bias seen for EDTA plasma samples. • No statistically significant difference was observed for serum samples drawn in the summer vs. winter (p≥0.418). Therefore, both seasons were combined and used to calculate the reference interval (Table 3).• The lower limit of the reference interval is not affected by vitamin D status for any method. However, the upper limit does seem to be influenced by the vitamin D cutoff used, except for the LIAISON. In most cases, the upper limit for vitamin D >30 ng/mL does not fall within the confidence limits of the calculated reference interval for all normal subjects as well as those with vitamin D >10 ng/mL. • While it remains controversial regarding the threshold used to define vitamin D insufficiency, some have reported using a vitamin D of 20 ng/mL. When analyzing our reference interval data at this cutoff, the upper reference limit for the majority of the methods, overlapped the 90% confidence intervals for the other cutoffs we looked at.

References1. Joly D, Drueke TB, Alberti C, Houillier P, Lawson-Body E, Martin KJ, et al. Variation in serum and plasma PTH levels in second-generation assays in hemodialysis patients: a cross-sectional study. Am J Kidney Dis 2008;51:987-95.2. Lips P. Vitamin D deficiency and secondary hyperparathyroidism in the elderly: consequences for bone loss and fractures and therapeutic implications. Endocr Rev 2001;22:477-501. 3. Malabanan A, Veronikis I, Holick M. Redefining vitamin D insufficiency. Lancet 1998;351:805-6.4. Souberbielle JC, Fayol V, Sault C, Lawson-Body E, Kahan A, Cormier C. Assay-specific decision limits for two new automated parathyroid hormone and 25-hydroxyvitamin D assays.5. Viljoen A, Singh DK, Twomey PJ, Farrington K. Analytical quality goals for parathyroid hormone based on biological variation. Clin Chem Lab Med 2008;46:1438-42.

Performance Characteristics of Six Intact Parathyroid Hormone AssaysSonia L. La’ulu1, William L. Roberts2

1ARUP Institute for Clinical and Experimental Pathology, Salt Lake City, UT, 2Department of Pathology, University of Utah, Salt Lake City, UTAmerican Association for Clinical Chemistry Annual Meeting, Chicago, Illinois. July 19-23, 2009

Abstract (revised)OBJECTIVE: The aim of this study was to evaluate the performance characteristics of six intact parathyroid hormone (PTH) assays: Beckman Coulter Access, Abbott ARCHITECT i2000SR, Siemens ADVIA Centaur, Roche Modular E170, Siemens IMMULITE 2000, and DiaSorin LIAISON. METHODS: Imprecision studies were performed using three concentrations of commercially available quality control materials. Two runs of duplicate testing were conducted per day, for 5 days, with a minimum of 2 hours between runs. For method comparison, 203 serum and 193 EDTA plasma samples were tested by all methods and analyzed using the E170 as the comparison method. To determine the reference interval, serum samples were drawn from 130 subjects in the summer (August) and winter (February) and a frozen aliquot was tested by all methods. A study comparing sample types and their stabilities was conducted where 15 individuals had 5 types of BD Vacutainer tubes drawn: red top, SST, PST, green top, and EDTA. An aliquot from each tube was tested after being subjected to one of the following conditions: tested immediately, frozen immediately and tested after thawing, and after 24 hours and 48 hours refrigerated. RESULTS: For imprecision, total CV’s ranged from 1.6 to 10.9%, 1.5 to 8.1%, and 1.1 to 5.5% for levels 1, 2, and 3, respectively. The E170 always had the lowest within laboratory CV and the LIAISON had the highest. Method comparison results by Passing-Bablok regression had slopes of 0.98 – 1.83 and correlation coefficients of 0.92 – 1.00. The lower limit of the reference interval ranged from 10.8 to 21.8 and the upper limit was 56.7 to 123.4. The results of the stability study are summarized (Table 4).CONCLUSIONS: In summary, imprecision was acceptable for all methods with total CV’s ≤10.9%. All methods correlated well to the E170 (r = 0.92 to 1.00). The IMMULITE exhibited the highest positive bias. The lower limit of the reference interval is not influenced by vitamin D status while the upper reference limit is affected. The types of collection tubes and sample storage conditions are more important for some methods than others.

IntroductionNumerous studies have demonstrated the lack of comparability among PTH assays. In addition, many note that the differences observed may potentially influence clinical decision making and prove to be problematic for subsequent diagnosis and treatment. This variability can be influenced by a variety of conditions such as the assay used, the population evaluated, vitamin D status, and numerous pre-analytical conditions. In the present study, we set out to evaluate performance characteristics that potentially play a role in inter-method variability of PTH assays namely, imprecision, comparison of methods including two sample types, reference intervals with consideration of vitamin D status, sample storage conditions, and sample collection tubes.

Materials and Methods• Testing was performed with six PTH assays: Beckman Coulter Access, Abbott ARCHITECT i2000SR, Siemens ADVIA Centaur, Roche Modular E170, Siemens IMMULITE 2000, and DiaSorin LIAISON. Testing by all methods was performed according to manufacturers’ instructions.• Imprecision studies were performed using three concentrations of commercially available quality control materials. Two runs of duplicate testing were conducted per day, for 5 days, with a minimum of 2 hours between runs. • Method comparison was evaluated by testing 203 serum and 193 EDTA plasma samples by all methods. Results were analyzed by Passing-Bablok regression with the E170 arbitrarily chosen as the comparison method.• To determine the reference interval, serum samples were drawn from 130 subjects in the summer and winter seasons and a frozen aliquot was tested by all methods. Samples were also tested for calcium, creatinine, and 25-hydroxyvitamin D (vitamin D). Those with an eGFR ≥60 and normocalcemic were used to calculate the reference interval. Reference interval results were further analyzed based on vitamin D results using cutoffs of 10, 20, and 30 ng/mL.• A study comparing sample types and their stabilities was conducted where 15 individuals had 5 types of BD Vacutainer tubes drawn: red top, SST, PST, green top, and EDTA. An aliquot from each tube was tested after being subjected to one of the following conditions: tested immediately (fresh), frozen immediately and tested after thawing (frozen), and after 24 hours and 48 hours refrigerated.• Results from samples that were frozen or refrigerated were compared to fresh samples using paired t-test and percent recoveries. Also, to evaluate possible differences among sample types, all collection tubes were compared to red tops for fresh samples.

AcknowledgementsSupport for this study was provided by Abbott Diagnostics and the ARUP Institute for Clinical and Experimental Pathology. We gratefully acknowledge Abbott Diagnostics, Beckman Coulter, and Roche Diagnostics for providing instrumentation to perform testing using their methods.

% CV

Method QCMean

concentration (pg/mL)

Within Run

Between Day

Total

Access Level 1 44.5 3.2 3.7 4.9

Level 2 264.8 2.0 2.2 3.4

Level 3 810.3 2.5 2.7 3.6

ARCHITECT Level 1 85.0 3.9 < 0.1 4.1

Level 2 451.4 2.9 < 0.1 3.7

Level 3 1472.1 3.0 0.8 3.1

Centaur Level 1 28.8 5.4 5.0 7.4

Level 2 188.2 2.1 1.4 2.5

Level 3 608.3 2.2 2.2 3.1

E170 Level 1 32.8 0.7 1.4 1.6

Level 2 196.9 1.2 0.9 1.5

Level 3 594.5 0.7 0.8 1.1

IMMULITE Level 1 41.3 5.8 2.7 6.4

Level 2 344.6 6.0 1.7 6.2

Level 3 1052.5 3.7 < 0.1 3.7

LIAISON Level 1 38.3 2.1 10.0 10.9

Level 2 205.9 1.5 7.9 8.1

Level 3 621.3 1.0 5.1 5.5

Method Sample type N R Slope Intercept SMAD‡

Access EDTA Plasma 193 0.994 0.98 -6.58 5.9

Serum 203 0.974 1.01 -2.99 10.9

ARCHITECT EDTA Plasma 193 0.998 1.27 -3.49 4.5

Serum 203 0.994 1.22 -1.05 10.1

Centaur EDTA Plasma 193 0.994 1.14 -7.49 7.0

Serum 203 0.966 1.07 -8.61 20.1

IMMULITE EDTA Plasma 193 0.991 1.83 -11.68 12.0

Serum 203 0.985 1.32 -6.04 24.5

LIAISON EDTA Plasma 193 0.986 1.16 -2.35 11.5

Serum 203 0.918 1.09 16.98 41.0

MethodFresh mean

Frozen mean

p value

24 hour refrigerated

p value

48 hour refrigerated

p value

Red top

Access 35.4 35.5 0.904 32.9 0.001 32.9 0.003

ARCHITECT 44.6 43.5 0.022 40.9<0.00

140.1 <0.001

Centaur 38.7 36.6 0.057 35.0<0.00

134.5 <0.001

E170 40.5 38.9 <0.001 37.7<0.00

136.6 <0.001

IMMULITE 45.9 44.1 0.031 42.7 0.001 41.6 <0.001

LIAISON 57.7 58.8 0.025 56.3 0.083 53.8 0.004

SST

Access 34.0 33.8 0.432 30.9<0.00

129.7† <0.001

ARCHITECT 42.3 42.0 0.452 38.3<0.00

138.0 <0.001

Centaur 35.0 35.2 0.447 31.4<0.00

129.9† <0.001

E170 37.4 36.9 0.016 34.7<0.00

133.8 <0.001

IMMULITE 43.2 40.2 0.006 39.3 0.001 38.6 0.001

LIAISON 56.0 56.7 0.307 52.7 0.003 51.1 0.001

PST

Access 36.4 35.3 0.159 35.9 0.345 34.6 0.011

ARCHITECT 45.5 42.2 0.002 43.9 0.047 42.9 0.001

Centaur 38.0 34.0 <0.001 36.2 0.034 33.3 <0.001

E170 43.5 39.7 <0.001 41.5 0.002 40.1 <0.001

IMMULITE 58.6* 50.3† <0.001 53.1 0.004 50.9† <0.001

LIAISON 56.6 52.3 0.013 56.4 0.938 53.5 0.042

Green top

Access 37.9 36.5 0.038 37.0 0.158 35.6 0.003

ARCHITECT 46.8 42.8 <0.001 44.6 0.011 44.0 0.002

Centaur 40.4 34.9† <0.001 37.7 0.005 37.2 0.008

E170 44.9 40.5 <0.001 42.2<0.00

141.5 <0.001

IMMULITE 59.0* 52.5 <0.001 53.9 0.006 54.1 0.006

LIAISON 61.8 53.3† 0.013 56.9 0.198 55.2 0.079

EDTA

Access 38.3 37.3 0.193 36.4 0.007 37.0 0.079

ARCHITECT 43.2 45.3 0.001 46.6<0.00

147.5 0.018

Centaur 43.4 40.7 0.029 42.5 0.142 42.6 0.245

E170 46.0* 41.4 <0.001 43.5<0.00

143.4 <0.001

IMMULITE 76.8* 70.6 0.002 75.5 0.307 72.1 0.001

LIAISON 57.3 51.7 0.158 52.0 0.200 50.9 0.108

• Analysis of sample stability showed many statistically significant differences compared to fresh analysis. However, when using a previously defined desirable analytical bias of 12.5% to evaluate if the differences where clinically significant, only 6 were ≥12.5% (Table 4). • No clinically significant difference was observed with any method for storage at 24 hours refrigerated vs. tested immediately.• Evaluation of changes among the different collection tubes revealed 4 which had a significant difference compared red top: PST, green top, and EDTA for IMMULITE and EDTA for the E170.• The types of collection tubes and sample storage conditions are more important for some methods than others.

y = 0.98x -6.58r = 0.994

0

250

500

750

1000

1250

1500

0 250 500 750 1000 1250 1500

E170 plasma

Ac

ce

ss

pla

sm

a

y = 1.27x - 3.49r = 0.998

0

250

500

750

1000

1250

1500

0 250 500 750 1000 1250 1500

E170 plasma

AR

CH

ITE

CT

pla

sm

a -

y = 1.14x -7.49r = 0.994

0

250

500

750

1000

1250

1500

0 250 500 750 1000 1250 1500

E170 plasma

Ce

nta

ur

pla

sm

a

y = 1.83x - 11.68r = 0.991

0

250

500

750

1000

1250

1500

0 250 500 750 1000 1250 1500

E170 plasma

IMM

UL

ITE

pla

sm

a

y = 1.16x - 2.35r = 0.986

0

250

500

750

1000

1250

1500

0 250 500 750 1000 1250 1500

E170 plasma

LIA

ISO

N p

las

ma

y = 1.01x - 2.99r = 0.974

0

500

1000

1500

2000

2500

0 500 1000 1500 2000 2500

E170 serum

Ac

ce

ss

se

rum

y = 1.22x - 1.05r = 0.994

0

500

1000

1500

2000

2500

0 500 1000 1500 2000 2500

E170 serum

AR

CH

ITE

CT

se

rum

-

y = 1.07x - 8.61r = 0.966

0

500

1000

1500

2000

2500

0 500 1000 1500 2000 2500

E170 serum

Ce

nta

ur

se

rum

y = 1.32x - 6.04r = 0.985

0

500

1000

1500

2000

2500

0 500 1000 1500 2000 2500

E170 serum

IMM

UL

ITE

se

rum

y = 1.09x + 16.98r = 0.918

0

500

1000

1500

2000

2500

0 500 1000 1500 2000 2500

E170 serum

LIA

ISO

N s

eru

m

Method N Lower limit (90% CI) Upper limit (90% CI)

Access 252 13.7 (11.8 – 15.1) 77.2 (67.0 – 89.3)

Vitamin D ≥10 243 13.6 (11.2 – 14.9) 74.9 (64.6 – 109.9)

Vitamin D ≥20 198 13.6 (11.8 – 14.9) 67.2 (57.0 – 75.0)

Vitamin D ≥30 133 13.7 (12.6 – 15.1) 62.5 (54.5 – 75.0)

ARCHITECT 252 18.1 (14.1 – 19.1) 88.5 (77.2 – 102.4)

Vitamin D ≥10 243 18.1 (13.7 – 19.0) 84.2 (76.2 – 123.9)

Vitamin D ≥20 198 18.1 (13.7 – 19.0) 80.4 (67.6 – 100.3)

Vitamin D ≥30 133 18.1 (15.6 – 19.1) 74.7 (62.3 – 81.9)

Centaur 252 11.0 (10.5 – 12.8) 70.6 (64.0 – 80.9)

Vitamin D ≥10 243 10.9 (10.2 – 12.6) 69.1 (62.7 – 85.0)

Vitamin D ≥20 198 10.8 (10.2 – 12.6) 62.7 (55.2 – 70.6)

Vitamin D ≥30 133 11.6 (10.2 – 12.8) 56.7 (50.9 – 70.6)

E170 252 17.3 (16.6 – 18.9) 73.5 (66.5 – 83.3)

Vitamin D ≥10 243 17.1 (14.6 – 18.7) 72.9 (64.4 – 95.6)

Vitamin D ≥20 198 17.1 (16.6 – 18.9) 64.5 (58.2 – 74.8)

Vitamin D ≥30 133 17.8 (16.8 – 18.9) 60.3 (51.3 – 74.8)

IMMULITE 252 13.8 (10.8 – 15.7) 92.2 (81.1 – 124.0)

Vitamin D ≥10 243 13.7 (7.0 – 14.7) 85.1 (79.0 – 150.0)

Vitamin D ≥20 198 13.9 (10.8 – 15.8) 79.1 (66.2 – 112.0)

Vitamin D ≥30 133 13.9 (10.8 – 16.8) 74.6 (61.3 – 84.1)

LIAISON 252 21.8 (17.1 – 25.0) 123.4 (108.0 – 153.0)

Vitamin D ≥10 243 21.8 (15.5-23.7) 119.5 (105.0 – 167.0)

Vitamin D ≥20 198 19.9 (15.5 – 25.2) 115.1 (103.0 – 129.0)

Vitamin D ≥30 133 18.0 (13.6 – 25.9) 109.9 (93.6 – 129.0)

*Greater than 12.5% desirable analytical bias than compared to red top.

†Greater than 12.5% desirable analytical bias than compared to samples tested immediately (fresh).

Table 1. Imprecision

Figure 1. Method comparison by Passing-Bablok regression for plasma and serum samples

Table 2. Method comparison Table 3. Reference interval Table 4. Sample stability and tube type