- Email: [email protected]
Serum human chorionic gonadotrophin levels in early pregnancy
Chnicu Chimicn Actn, 181 (1989) 281-292 Elsevier
281
CCA 04432
Serum human chorionic gonadotrophin in early pregnancy Daylily
S. Ooi ‘y2, Sherry
L. Perkins
‘, Paul
Claman
3 and Henry
levels F. Muggah
t Department ofLaboratoryMedicine, Ottawa Civic Hospital, ’ Departments of Biochemistry Pathology, University of Ottawa and 3 Department of Obstetrics and Gynecology, Ottawa Civic Hospitol and University of Ottawa, Ottawa, Ontario (Canada)
3
and
(Received 28 June 1988; revision received 9 December 1988; accepted 27 December 1988) Key worak Serum human chorionic gonadotrophin levels; Pregnancy
Summary
Serum hCG reference intervals for various gestational periods in normal pregnancies were determined using three commercial assays - two standardized against the WHO 2nd IS (Amersham Amerlex-M j3HCG RIA (AMX) and Abbott P-HCG 15/15 (ABB)) and one standardized against the WHO 1 IRP (Hybritech Tandem @-E HCG (HYB)). Serial samples from patients with accurately determined gestational periods were analyzed. We correlated these assays to determine the validity of the common practice of interchanging values between assays using the same WHO standard and of converting 1st IRP values to 2nd IS values by a fixed factor. The slope of correlation between the two 2nd IS assays (AMX, ABB) was 1.43, r = 0.960; whereas between the 1 IRP assay (HYB) and the two 2nd IS assays the slopes were 1.67, r = 0.963 and 1.22, r = 0.971 for AMX and ABB, respectively. In a prospective study of 52 patients with normal pregnancies, serum beta-hCG values in 46% of samples taken at 28-35 days gestation fell below the lower limit of the reference curves supplied with the AMX kit. Ninety-two percent of samples were within the newly established intervals. These results indicate that supplier’s reference limits may not be accurate; in addition, a common factor should not be used to convert values from one commercial kit to another.
Introduction
The measurement of human chorionic gonadotrophin (hCG) in serum and urine is now an established method for confirmation of pregnancy. Clinically it is most
Correspondence to: D.S. Ooi, Medical Biochemist, Ottawa Civic Hospital, 1053 Carling Avenue, Ottawa, Ontario, Canada KlY 4E9.
0009~8981/89/$03.50
0 1989 Elsevier Science Publishers B.V. (Biomedical Division)
282
useful in the early diagnosis of ectopic pregnancies. Commonly used non-invasive investigations for ectopic pregnancy are serum hCG levels and ultrasonography. We [l] and other investigators [2-41 have shown that, using currently available serum hCG assays, almost all patients with ectopic pregnancies have detectable levels at time of clinical presentation. Since sonography is usually not helpful before the sixth week of gestation [4], serum hCG levels are the only diagnostic tool useful in this early period. Serum hCG levels are used in several ways. Single values may be correlated with gestational age. However this is dependent on accurate menstrual history and correct reference intervals for each period of gestation. In addition, because of the rapid rise in hCG levels during early pregnancy, reference intervals have very wide ranges. The other important use of a single hCG value is in sonography. Kadar et al. [5] advocate the use of a ‘discriminatory hCG zone’ to aid in the diagnosis of ectopic pregnancies. In their series, serum hCG levels above its upper limit are consistently associated with a gestational sac in normal pregnancies, whereas below the lower limit, the sac is not expected to be visualized. Other reports of the use of this ‘zone’ [6-81 have since confirmed its usefulness. However, differences in ultrasound technique and hCG assays have resulted in considerable confusion concerning the numerical value of the upper limit of this zone. When serial serum hCG values are available, one can assess the rate of increase. This has been expressed as percent increase [5], the slope of the log hCG versus time 191, and doubling time [lo-121. However, these approaches have their limitations, as will be discussed later. We have found plotting serial levels against serum hCG reference curves to be a convenient way of observing the rise. This also allows interpretation of each value against the reference intervals. Valid reference curves are essential for this. The reference intervals for three commercial assays were established using serial samples from normal pregnancies with reliable gestational dates. We also examined the correlation of the assay used in our laboratory with two other assays.
Methods and materials Our laboratory routinely uses the Amerlex-M PHCG kit (Amersham plc, Amersham, Bucks, UK). Two other commercial kits were evaluated, the Tandem @-E HCG ImmunoEnzyMetric Assay (Hybritech - Inc., San Diego, CA, USA) and the Abbott P-HCG 15/15 (Abbott Labs., North Chicago, IL, USA). The AMX and ABB kits are standardized against the World Health Organization (WHO) Second International Standard (2 IS) while the HYB kit is standardized against the WHO First Reference Preparation (1 IRP). The assays were performed according to the manufacturers’ instructions. Dilutions were made using diluent supplied in the kit to 3, 15, 75, 375 or 750 times original sample as required. The precision of assays were monitored using Lyphochek” Human Control Serum master lot 30000 (BioRad, Anaheim, CA). In order to assess consistency of results we obtained with HYB and ABB kits with other laboratories who routinely
283
use these kits, we compared our results of the above quality control with manufacturer’s published summaries of eight other ABB and nine other HYB users. Serial blood samples were collected in Vacutainer TMtubes (Becton Dickinson, serum was separated and stored at Mississauga, Canada) without preservative; - 20° C until analysis. Insufficient sample volumes prohibited analysis of all samples using each of the three commercial kits. The AMX kit was evaluated with 212 samples from 60 patients who progressed to normal pregnancies; 41 of whom had three or more samples, with a mean of 4.1 samples per patient. Reference intervals for HYB were established by assaying a total of 193 samples from 55 of these patients. 157 samp!es from 57 patients were assayed using the ABB kit. Gestational dating was based on one or more of the following criteria: basal body temperature shift, embryo transfer after in vitro fertilization, or last menstrual period data supported by first trimester ultrasound findings. We calculated reference intervals using 4-day periods. As the values had a log-normal distribution within each period, reference intervals were determined following log transformation. To evaluate the reference intervals obtained with the AMX assay, we prospectively studied a further group of 91 patients, 52 (77 samples) normal pregnancies with gestations dated as in the study group, 24 (45 samples) ectopic pregnancies and 15 (46 samples) abortions with serum samples taken between 28 and 52 days from last menstrual period. Doubling times were calculated according to the formula: DT = 0.3t/log,,(c,,‘c,), where DT = time in days for serum hCG levels to double, concentration, c2 = the second serum hCG concentration, serum hCG samples.
c1 = the first serum hCG t = days between the two
Results
Precision The between run coefficient of variation of the three assay kits was determined using Lyphochek Human Control Sera. All three assays had acceptable interassay precision (Table I). No significant difference in assay precision was noted by the F test ( p > 0.05) except for AMX against the other two assays for Lyphochek II.
Reference interuds During the first 40 days of gestation, serum hCG levels increased at a constant exponential rate. Thereafter the rate of rise decreased with levels in most patients plateauing by day 65 (Tables II-IV). Values obtained were compared with the reference intervals provided in the kit inserts (Figs. l-3).
284 TABLE
I
Between run precision
Lyphochek I
n Mean (IU/l)
Lyphochek II
CV (W) ” Mean (II-J/I)
Lyphochek III
CV (W) n Mean (IU/l) CV (W)
TABLE
AMX
HYB
ABB
53 6.74 13.1 53 26.9 6.4 46 147 9.8
11 13.42 10.3 11 34.4 10.8 9 162 6.7
15 8.91 14.0 14 32.3 14.9 12 160 11.7
II
Reference intervals using Amerlex-M BHCG Days gest a
2832364044 4852566070-
”
-ii31 34 34 34 18 15 11 12 10
RIA
S hCG (IU/I 2 IS) Mean
Lower limit b
Upper limit b
100 640 2400 6800 14000 24000 38000 53000 50000 41000
20 60 320 1100 3900 8700 19000 25000 24000 14000
590 6400 19000 43000 49000 67000 79000 110000 100000 120000
a From first day of last menstrual period. b Determined using log-transformed values. See text for details. TABLE
III
Reference intervals using Hybritech” Davs nest ’ 2
”
n
Tandem-E HCG S hCG W/l .I
Mean 28-
13
200
32364044 4852566070-
26 33 31 33 16 14 11 10 9
1100 4100 llooo 25000 41000 64000 83000 100000 75000
IRP)
I
Lower limit b
Upper limit b
40 140 540 2100 6700 15000 24000 38000 51000 32000
1000 8600 31000 61000 93000 120000 170000 180000 200000 170000
’ From fist day of last menstrual period. b Determined using log transformed values. See text for details.
285 TABLE IV Reference intervals using Abbott 15/15 /3HCG Day gest ’
2832364044485256” 6070-
n
S hCG (W/l 2 IS)
12 24 25 27 16 13 14 8 7 8
Mean
Lower limit b
Upper limit b
120 870 3600 9200 20000 38OW 55000 60000 81000 43000
20 110 420 1800 6100 19000 24ooO 32W 48000 18000
690 7000 32000 46ooo 67000 74000 130000 1lOOtXl 140000 100000
a From first day of last menstrual period. b Determined using log transformed values. See text for details.
Evaluation of the manufacturer’s reference interval for the AMX kit showed 43% of the samples from between 28-35 days gestation and 10% of samples from between 36-52 days gestation would have fallen below the lower reference limit. In the prospective study group, for patients presenting before 35 days gestation, the
28
35
42
DAYS GESTATION
49
(FROM
56
63
LMP)
Fig. 1. Reference intervals established using Amerlex-M j?HCG RIA p. k&insert------.
Values as obtained from
286
DAYS GESTATION
Fig. 2. Reference
intervals
established
using Hybritech
(FROM
LMP)
Tandem-E
PHCG
-.
Values as obtained
from kit insert - - - - - -,
DAYS GESTATbON (FROM
Fig. 3. Reference
intervats
established
LMP)
using Abbott &HCG 15/15 insert - - - - - -.
---.
Values as obtained
from kit
287 TABLE
V
Correlation
between assays
S hCG (IU/I 2 IS)
Slope
n
Ratio
r
Mean
(SD)
HYB vs AMX 78 111 189
o-5 999 6~-1~~ All
1.71 1.67 1.67
0.975 0.937 0.963
1.79 1.75 1.77
(0.30) (0.36) (0.36)
1.37 1.43 1.43
0.967 0.932 0.960
1.34 1.42 1.39
(0.25) (0.27) (0.28)
1.45 1.22 1.22
0.961 0.957 0.971
1.40 1.26 1.31
(0.30) (0.28) (0.29)
ABB vs. AMX
65 95 160
o-5 999 6000-loOoM3 All
HYB vs ABB 52 107 159
o-5 999 6~-1~~ All
manufacturer’s reference curves ~sclassified 46%, 20% and 21% of normal, ectopic and aborting pregnancies respectively. Using our newly established reference limits, the corresponding percentages were 8, 20 and 49. .lt1
Yl
Y2
+
7.0
t
0
+
0
6.0
# l
+
+
8
24
5.0
+ +
pJ 4.0
0
u
a, m m
z i
*lo
3.0
8 cw co
2.0
1.0
cl 0.
mu.
+
v 0 *
3 # +
6000.
;
+
f
0
+
*c *
$
0
*
cl
+ +
li%ii-
3OcQo-
hCG (W/L2 IS)
Fig. 4. Doubling times of serum hCG, samples taken 2-4 days to), and 5-8 days (+ ) apart at various stages of gestation. (*) represents DT values > 7 days, with number of occurrences next to it. The bars represent reference intervals established using log cumulative frequency.
288
Correlation Relationships between the assays were studied by linear regression analysis by arithmetic ratios. The slopes obtained with linear regression, forcing the through zero, were similar to arithmetic ratios for all three correlations (Table The ratio of values obtained with the two 2 IS assays, ABB to AMX, 1.39 + 0.28 (mean + SD) where the slope was 1.43, r = 0.960. Between the 1 assay (HYB) and the 2 IS assays, the ratios were 1.77 + 0.3 (AMX) and 1.31 f (ABB) while the corresponding slopes were 1.67 and 1.22.
and lines V). was IRP 0.29
Doubling time (DT) Doubling times of hCG levels taken two to four days (n = 68 pairs) and 5-8 days (n = 104 pairs) apart were calculated and grouped according to the second hCG level. As the distribution of DTs showed a log normal distribution, reference intervals were established using log probability paper as the 2.5 and 97.5 cumulative frequency values [13] (Fig. 4). Using the Student’s t test, there were no significant differences between the first three groups (hCG values between 0 and 17999 IU/l); for samples between 18 000 and 29999 IU/l the doubling times were significantly different from the first group. Discussion Reference curves established using our patients’ data showed the exponential increase seen previously by many investigators. However, the reference curves for the three assays differed from each other as well as from those supplied by the manufacturers. These differences did not arise from variations in technique as the mean values for the three levels of Lyphochek Human control sera from our laboratory were within one standard deviation of the group means from the QC manufacturer’s published summaries of other users. Of the three, the reference curve obtained for HYB was most similar to the manufacturer’s. With ABB, values were markedly different since the manufacturer’s values were not obtained using the assay itself. For AMX, there was discrepancy in the early part of the curve, which may be partly attributable to the manufacturer plotting upper and lower limit values at the beginning rather than the middle of each interval thus shifting the curve to the left. This difference is clinically significant, as up to 46% of samples from normal pregnancies taken between 28-35 days gestation could be interpreted as abnormally low using the manufacturer’s reference intervals. Shifting the curve to the right overcame this, but resulted in poorer detection rate of abnormal pregSince abortions are not usually life nancies, mainly in patients with abortions. threatening, this misclassification is generally not as significant as misdiagnosing a normal pregnancy as abnormal. Variations in reference values for different assays kits stem not only from the use of two WHO standards giving numerically dissimilar results but reported in the same units (IU/l), but from assay differences as well. The WHO 2 IS was prepared in 1963 for bioassays, and the 1 IRP or HCG 75/589 was prepared in 1975 for immunoassays. It is commonly accepted that values obtained with 1 IRP assays are 1.8 [14] to 2.07 [15] times that of 2 IS assay values. Lack of understanding of the
289
relationship between these WHO standards has led to confusion concerning the value of the ‘discriminatory zone’. This is exemplified by the fact that Kadar et al. [16] originally published their data for the “discriminatory zone” for serum hCG as 6000-6 500 IU/l 2nd IRP. This standard does not exist, but this value has subsequently been cited in the literature as applicable to both 2 IS and 1 IRP. Batzer et al. [7] in a prospective evaluation of 146 ultrasound observations on 98 pregnancies established 1000 ng/ml (or 5 600 IU/l 2 IS) as the upper limit. In a large prospective study, Kadar’s group [5] confirmed their previous findings, but stated that the hCG assay was calibrated against 1 IRP. Of 36 patients with ectopic pregnancies, 86% showed no sac with hCG levels above 6500 IU/l 1 IRP (or 3250 IU/l 2 IS). These authors erroneously concluded that their results were similar to those of Batzer, even though they were actually stating a value approximately half that of Batzer’s. In 1985, Nyberg [17], using a 2 IS hCG assay, reported a level of 1800 IU/l above which, the absence of an intrauterine sac was 100% sensitive for abnormal pregnancy. Unfortunately these studies have been widely quoted and misinterpreted with regard to standardization resulting in confusion and no clear consensus as to what the ‘discriminatory zone’ should be. Further complicating this issue is the considerable variability in sensitivity with ultrasound techniques. In addition to these problems of standardization, differences in hCG kits must also be considered. Variability in hCG values even when standardized similarly was reported as far back as 1983 when Rasor and co-workers [18] showed variable correlations between the National Institute of Health kit, three 1 IRP and seven 2 IS kits. They could not assign a relationship between the two WHO standards. The differences in values obtained using kits standardized similarly have been attributed to cross-reactivity with other gonadotrophins, and differences in hCG components detected. Commercial kits are now beta-specific to avoid cross reactivity with other pituitary glycoprotein hormones. However, considerable kit to kit variability may arise from the use of monoclonal versus polyclonal antisera as well as methodology. Of the three kits evaluated in this study, AMX uses a polyclonal antiserum which measures dimer hCG (intact (Y/P), free /?-subunits and fragments containing P-subunits. The ABB uses polyclonal antiserum, while HYB is monoclonal, and both measure only intact a/j3 dimer hCG. Hay [19] has shown that this variation in ligand specificity may yield discordant results when comparing assays. He postulated that discrepancies diminish with increasing gestational age. The immature trophoblast synthesizes predominantly free P-subunits; while mature cytotrophoblast produces sufficient a-subunits to match the P-subunits and yield mainly whole hCG dimer. There are also eight alleles for the P-subunit, so that even P-subunit specific assays may vary depending on whether monoclonal or polyclonal antiserum is used. This is often not considered when interpreting results involving serum hCG as long as kits are standardized against the same WHO standard. Recent papers comparing methods using the same standards, ABB versus AMX [20] and Stratus fluoroimmunoassay VS. HYB [21], have concluded that minor differences between the assays studied are analytically acceptable. Our study has shown that, with the kits evaluated, this may not be true for all kits. Values obtained with the ABB were on the average 1.4 times the value of
290
AMX. Both assays are standardized against 2 IS. The ratios between HYB which is standardized against 1 IRP and AMX and ABB were 1.8 and 1.3, respectively. The common practice of using published data generated by different assays [22-251 should be discouraged. A conversion factor may be used when a specific relationship between the two assays is known. However, caution should be exercised with results obtained using diluted samples, since matrix effects of dilution may make correlations nonlinear in the higher range. Using the rate of increase rather than absolute values may circumvent problems of assay variation and inaccurate gestational dates. However, considerable controversy has arisen as to the proper determination and interpretation of these values. Batzer et al. [9] obtained a DT of 2.2 + 1.0 days from the slope of the linear regression analysis of log hCG vs. gestational age during the first 44 days. Kadar et al. [16] determined the normal increase in serum hCG over 48 hours to be > 66%. In 1985, Pittaway et al. [lo] demonstrated that: (a) the exponential rate of increase in serum hCG decreases with gestational age and is more accurately represented by a quadratic equation, and (b) the DT of serum hCG increases with increasing hCG level or gestational age. They subsequently [26] showed that use of multiple values of DTs was more sensitive than use of a single mean DT determined by linear regression. These findings were confirmed by Daya [ll] who established values for three separate gestational periods, and Fritz [12] who proposed the use of a nomogram based on the level of hCG and the time between samples. This representation of hCG increase with gestational age has been criticized [27] as has the contention that DTs vary with gestational age, sampling interval and hCG. Kadar and Romero [28] have recently found that estimates of DT for serum hCG are not influenced by initial hCG, gestational age or sampling interval. Our data is consistent with a change in DT with gestational age; beyond 38 days, or 6000 IU/l 2 IS, DTs were unpredictable and of little use clinically. Furthermore, the use of DT is not always feasible. For patients with ectopic pregnancy, it may not be prudent to wait 48 h. In patients with abortion, samples taken more than 4 days previous may not be useful for determination of DT, as normal increases prior to the embryonal death may mask the fall off in hCG levels. In practice, we find evaluation of abnormal pregnancies with either doubling time or hCG value against reference interval alone inadequate in many patients. Plotting available values on a reference curve combines both methods, allowing us to evaluate each result against the reference interval as well as visually compare the observed rate of increase with the expected. Valid reference intervals are essential for this approach. Ideally each laboratory should establish its own reference intervals since even those determined by the manufacturer may not be appropriate. Unfortunately, with the rapid rise in levels in early pregnancy, a large number of patient samples is required to establish reliable reference intervals. It is imperative that clinicians be made aware of this variability in hCG assays and the pitfalls of using published data for interpretation of hCG values.
291
Acknowledgement The authors
thank
Amersham,
Hybritech
and Abbott
for supplying
kits.
References 1 Ooi DS, Perkins SL. Serum #IhCG assays in ectopic pregnancies (abstr.). Clin Chem 2 Cartwright PS, DiPietro DL. Ectopic pregnancy. Changes in serum human chorionic
1986;32:1166. gonadotropin
concen&ation., Obstet Gynecol 1984;63:76-80. 3 Braunstein GD, Asch RH. Predictive value analysis of measurements of human chorionic gonadotropin, pregnancy specific /3-l-glycoprotein, placental lactogen and cystine aminopeptidase for the diagnosis of ectopic pregnancy. Fertil Steril 1983;39:62-67. 4 Batzer FR, Corson SL. Diagnostic techniques used for ectopic pregnancy. J Reprod Med 1986;31:86-93. 5 Kadar N, Caldwell 6 7
8 9
10 11 12 13 14 15 16 17 18 19 20 21 22 23
BV, Romero
R. A method
of screening
for ectopic
pregnancy
and its indications.
Obstet Gynecol 1981;58:162-165. Romero R, Kadar N, Jeanty P, et al. Diagnosis of ectopic pregnancy: value of the discriminatory human chorionic gonadotropin zone. Obstet Gynecol 1985;66:357-360. Batzer FR, Weiner S, Corson SL, Schlaff S, Otis CV. Landmarks during the first forty-two days of gestation demonstrated by the /?-subunit of human chorionic gonadotropin and ultrasound. Am J Obstet Gynecol 1983;146:973-979. Nyberg DA, Filly RA, Filho DLD, Laing FC, Mahony BS. Abnormal pregnancy: early diagnosis by US and serum chorionic gonadotropin level. Radiology 1986;158:393-396. Batzer FR, Schlaff S, Goldfarb AF, Corson SL. Serial B-subunit human chorionic gonadotropin doubling time as a prognosticator of pregnancy outcome in an infertile population. Fertil Steril 1981;35:307-312. Pittaway DE, Reish RL, Colson Wentz A. Doubling times of human chorionic gonadotropin increase in early viable intrauterine pregnancies. Am J Obstet Gynecol 1985;152:299-302. Daya S. Human chorionic gonadotropin increase in normal early pregnancy. Am J Obstet Gynecol 1987;156:286-290. Fritz MA, Guo S. Doubling time of human chorionic gonadotropin (hCG) in early normal pregnancy: relationship to hCG concentration and gestational age. Feril Steril 1987;47:584-589. Neuman GJ. The determination of normal ranges from routine laboratory data. Clin Chem 1968;14:979-988. Rule AH, Michlewitz H, Boyle E, Donaboe M. Use of beta-human chorionic gonadotropin in gestational aging. Ann Clin Lab Sci 1985;15:428-434. Amerlex-M PHCG RIA kit insert. Amersham International plc, Amersham U.K. Kadar N, DeVore G, Romero R. Discriminatory hCG zone: its use in the sonographic evaluation for ectopic pregnancy. Obstet Gynecol 1981;58:156-161. Nyberg DA, Filly RA, Laing FC, Mack LA, Zarutskie PW. Ectopic pregnancy. Diagnosis by sonography correlated with quantitative HCG levels. J Ultrasound Med 1987;6:145-150. Rasor JL, Farber S, Braunstein GD. An evaluation of 10 kits for determination of human chorionic gonadotropin in serum. Clin Chem 1983;29:1828-1831. Hay DL. Discordant and variable production of human chorionic gonadotropin and its free (I- and P-subunits in early pregnancy. Clin Endocrinol Metab 1985;61:1195-1200. Hether NW, Gadsden RH. Evaluation of the Abbott p-hCG 15/15 enzyme immunoassay (EIA) kit with comparison to two other methods (Abst.). Clin Chem 1985;31:959. Beinlich CJ, Carpenter RH. Automated and manual quantitative assays of chorionic gonadotropin in serum compared. Clin Chem 1987;33:167-169. LeMaistre A, Bracey A, Katz A, Wu AHB. Role of qualitative chorio-gonadotropin assays in diagnosis of ectopic pregnancy. Clin Chem 1987;33:1908-1910. Hull ME, Schulman H. Impact of P-subunit human chorionic gonadotropin and ultrasonography in the management of suspected ectopic pregnancy. J Reprod Med 1986;31:478-482.
292 24 Pike Matthews C, Coulson PB, Wild RA. Serum progesterone levels as an aid in the diagnosis of ectopic pregnancy. Obstet Gynecol 1986;68:390-394. 25 Grudzinskas JG, Westergaard JG, Poulson H, Teisner B. The value of hCG measurements in the detection of ectopic pregnancy. Amersham International plc, Amersham U.K. 26 Pittaway DE, Colston Wentz A. Evaluation of early pregnancy by serial chorionic gonadotropin determinations: a comparison of methods by receiver operating characteristic curve analysis. Fertil Steril 1985;43:529-533. 27 Kadar N. Relationship between log of the human chorionic gonadotropin concentration and time period in early pregnancy [Letter]. Am J Obstet Gynecol 1986;154:692-693. 28 Kadar N, Romero R. Observations on the log human chorionic gonadotropin-time relationship in early pregnancy and its practical implications. Am J Obstet Gynecol 1987;157:73-78.
281
CCA 04432
Serum human chorionic gonadotrophin in early pregnancy Daylily
S. Ooi ‘y2, Sherry
L. Perkins
‘, Paul
Claman
3 and Henry
levels F. Muggah
t Department ofLaboratoryMedicine, Ottawa Civic Hospital, ’ Departments of Biochemistry Pathology, University of Ottawa and 3 Department of Obstetrics and Gynecology, Ottawa Civic Hospitol and University of Ottawa, Ottawa, Ontario (Canada)
3
and
(Received 28 June 1988; revision received 9 December 1988; accepted 27 December 1988) Key worak Serum human chorionic gonadotrophin levels; Pregnancy
Summary
Serum hCG reference intervals for various gestational periods in normal pregnancies were determined using three commercial assays - two standardized against the WHO 2nd IS (Amersham Amerlex-M j3HCG RIA (AMX) and Abbott P-HCG 15/15 (ABB)) and one standardized against the WHO 1 IRP (Hybritech Tandem @-E HCG (HYB)). Serial samples from patients with accurately determined gestational periods were analyzed. We correlated these assays to determine the validity of the common practice of interchanging values between assays using the same WHO standard and of converting 1st IRP values to 2nd IS values by a fixed factor. The slope of correlation between the two 2nd IS assays (AMX, ABB) was 1.43, r = 0.960; whereas between the 1 IRP assay (HYB) and the two 2nd IS assays the slopes were 1.67, r = 0.963 and 1.22, r = 0.971 for AMX and ABB, respectively. In a prospective study of 52 patients with normal pregnancies, serum beta-hCG values in 46% of samples taken at 28-35 days gestation fell below the lower limit of the reference curves supplied with the AMX kit. Ninety-two percent of samples were within the newly established intervals. These results indicate that supplier’s reference limits may not be accurate; in addition, a common factor should not be used to convert values from one commercial kit to another.
Introduction
The measurement of human chorionic gonadotrophin (hCG) in serum and urine is now an established method for confirmation of pregnancy. Clinically it is most
Correspondence to: D.S. Ooi, Medical Biochemist, Ottawa Civic Hospital, 1053 Carling Avenue, Ottawa, Ontario, Canada KlY 4E9.
0009~8981/89/$03.50
0 1989 Elsevier Science Publishers B.V. (Biomedical Division)
282
useful in the early diagnosis of ectopic pregnancies. Commonly used non-invasive investigations for ectopic pregnancy are serum hCG levels and ultrasonography. We [l] and other investigators [2-41 have shown that, using currently available serum hCG assays, almost all patients with ectopic pregnancies have detectable levels at time of clinical presentation. Since sonography is usually not helpful before the sixth week of gestation [4], serum hCG levels are the only diagnostic tool useful in this early period. Serum hCG levels are used in several ways. Single values may be correlated with gestational age. However this is dependent on accurate menstrual history and correct reference intervals for each period of gestation. In addition, because of the rapid rise in hCG levels during early pregnancy, reference intervals have very wide ranges. The other important use of a single hCG value is in sonography. Kadar et al. [5] advocate the use of a ‘discriminatory hCG zone’ to aid in the diagnosis of ectopic pregnancies. In their series, serum hCG levels above its upper limit are consistently associated with a gestational sac in normal pregnancies, whereas below the lower limit, the sac is not expected to be visualized. Other reports of the use of this ‘zone’ [6-81 have since confirmed its usefulness. However, differences in ultrasound technique and hCG assays have resulted in considerable confusion concerning the numerical value of the upper limit of this zone. When serial serum hCG values are available, one can assess the rate of increase. This has been expressed as percent increase [5], the slope of the log hCG versus time 191, and doubling time [lo-121. However, these approaches have their limitations, as will be discussed later. We have found plotting serial levels against serum hCG reference curves to be a convenient way of observing the rise. This also allows interpretation of each value against the reference intervals. Valid reference curves are essential for this. The reference intervals for three commercial assays were established using serial samples from normal pregnancies with reliable gestational dates. We also examined the correlation of the assay used in our laboratory with two other assays.
Methods and materials Our laboratory routinely uses the Amerlex-M PHCG kit (Amersham plc, Amersham, Bucks, UK). Two other commercial kits were evaluated, the Tandem @-E HCG ImmunoEnzyMetric Assay (Hybritech - Inc., San Diego, CA, USA) and the Abbott P-HCG 15/15 (Abbott Labs., North Chicago, IL, USA). The AMX and ABB kits are standardized against the World Health Organization (WHO) Second International Standard (2 IS) while the HYB kit is standardized against the WHO First Reference Preparation (1 IRP). The assays were performed according to the manufacturers’ instructions. Dilutions were made using diluent supplied in the kit to 3, 15, 75, 375 or 750 times original sample as required. The precision of assays were monitored using Lyphochek” Human Control Serum master lot 30000 (BioRad, Anaheim, CA). In order to assess consistency of results we obtained with HYB and ABB kits with other laboratories who routinely
283
use these kits, we compared our results of the above quality control with manufacturer’s published summaries of eight other ABB and nine other HYB users. Serial blood samples were collected in Vacutainer TMtubes (Becton Dickinson, serum was separated and stored at Mississauga, Canada) without preservative; - 20° C until analysis. Insufficient sample volumes prohibited analysis of all samples using each of the three commercial kits. The AMX kit was evaluated with 212 samples from 60 patients who progressed to normal pregnancies; 41 of whom had three or more samples, with a mean of 4.1 samples per patient. Reference intervals for HYB were established by assaying a total of 193 samples from 55 of these patients. 157 samp!es from 57 patients were assayed using the ABB kit. Gestational dating was based on one or more of the following criteria: basal body temperature shift, embryo transfer after in vitro fertilization, or last menstrual period data supported by first trimester ultrasound findings. We calculated reference intervals using 4-day periods. As the values had a log-normal distribution within each period, reference intervals were determined following log transformation. To evaluate the reference intervals obtained with the AMX assay, we prospectively studied a further group of 91 patients, 52 (77 samples) normal pregnancies with gestations dated as in the study group, 24 (45 samples) ectopic pregnancies and 15 (46 samples) abortions with serum samples taken between 28 and 52 days from last menstrual period. Doubling times were calculated according to the formula: DT = 0.3t/log,,(c,,‘c,), where DT = time in days for serum hCG levels to double, concentration, c2 = the second serum hCG concentration, serum hCG samples.
c1 = the first serum hCG t = days between the two
Results
Precision The between run coefficient of variation of the three assay kits was determined using Lyphochek Human Control Sera. All three assays had acceptable interassay precision (Table I). No significant difference in assay precision was noted by the F test ( p > 0.05) except for AMX against the other two assays for Lyphochek II.
Reference interuds During the first 40 days of gestation, serum hCG levels increased at a constant exponential rate. Thereafter the rate of rise decreased with levels in most patients plateauing by day 65 (Tables II-IV). Values obtained were compared with the reference intervals provided in the kit inserts (Figs. l-3).
284 TABLE
I
Between run precision
Lyphochek I
n Mean (IU/l)
Lyphochek II
CV (W) ” Mean (II-J/I)
Lyphochek III
CV (W) n Mean (IU/l) CV (W)
TABLE
AMX
HYB
ABB
53 6.74 13.1 53 26.9 6.4 46 147 9.8
11 13.42 10.3 11 34.4 10.8 9 162 6.7
15 8.91 14.0 14 32.3 14.9 12 160 11.7
II
Reference intervals using Amerlex-M BHCG Days gest a
2832364044 4852566070-
”
-ii31 34 34 34 18 15 11 12 10
RIA
S hCG (IU/I 2 IS) Mean
Lower limit b
Upper limit b
100 640 2400 6800 14000 24000 38000 53000 50000 41000
20 60 320 1100 3900 8700 19000 25000 24000 14000
590 6400 19000 43000 49000 67000 79000 110000 100000 120000
a From first day of last menstrual period. b Determined using log-transformed values. See text for details. TABLE
III
Reference intervals using Hybritech” Davs nest ’ 2
”
n
Tandem-E HCG S hCG W/l .I
Mean 28-
13
200
32364044 4852566070-
26 33 31 33 16 14 11 10 9
1100 4100 llooo 25000 41000 64000 83000 100000 75000
IRP)
I
Lower limit b
Upper limit b
40 140 540 2100 6700 15000 24000 38000 51000 32000
1000 8600 31000 61000 93000 120000 170000 180000 200000 170000
’ From fist day of last menstrual period. b Determined using log transformed values. See text for details.
285 TABLE IV Reference intervals using Abbott 15/15 /3HCG Day gest ’
2832364044485256” 6070-
n
S hCG (W/l 2 IS)
12 24 25 27 16 13 14 8 7 8
Mean
Lower limit b
Upper limit b
120 870 3600 9200 20000 38OW 55000 60000 81000 43000
20 110 420 1800 6100 19000 24ooO 32W 48000 18000
690 7000 32000 46ooo 67000 74000 130000 1lOOtXl 140000 100000
a From first day of last menstrual period. b Determined using log transformed values. See text for details.
Evaluation of the manufacturer’s reference interval for the AMX kit showed 43% of the samples from between 28-35 days gestation and 10% of samples from between 36-52 days gestation would have fallen below the lower reference limit. In the prospective study group, for patients presenting before 35 days gestation, the
28
35
42
DAYS GESTATION
49
(FROM
56
63
LMP)
Fig. 1. Reference intervals established using Amerlex-M j?HCG RIA p. k&insert------.
Values as obtained from
286
DAYS GESTATION
Fig. 2. Reference
intervals
established
using Hybritech
(FROM
LMP)
Tandem-E
PHCG
-.
Values as obtained
from kit insert - - - - - -,
DAYS GESTATbON (FROM
Fig. 3. Reference
intervats
established
LMP)
using Abbott &HCG 15/15 insert - - - - - -.
---.
Values as obtained
from kit
287 TABLE
V
Correlation
between assays
S hCG (IU/I 2 IS)
Slope
n
Ratio
r
Mean
(SD)
HYB vs AMX 78 111 189
o-5 999 6~-1~~ All
1.71 1.67 1.67
0.975 0.937 0.963
1.79 1.75 1.77
(0.30) (0.36) (0.36)
1.37 1.43 1.43
0.967 0.932 0.960
1.34 1.42 1.39
(0.25) (0.27) (0.28)
1.45 1.22 1.22
0.961 0.957 0.971
1.40 1.26 1.31
(0.30) (0.28) (0.29)
ABB vs. AMX
65 95 160
o-5 999 6000-loOoM3 All
HYB vs ABB 52 107 159
o-5 999 6~-1~~ All
manufacturer’s reference curves ~sclassified 46%, 20% and 21% of normal, ectopic and aborting pregnancies respectively. Using our newly established reference limits, the corresponding percentages were 8, 20 and 49. .lt1
Yl
Y2
+
7.0
t
0
+
0
6.0
# l
+
+
8
24
5.0
+ +
pJ 4.0
0
u
a, m m
z i
*lo
3.0
8 cw co
2.0
1.0
cl 0.
mu.
+
v 0 *
3 # +
6000.
;
+
f
0
+
*c *
$
0
*
cl
+ +
li%ii-
3OcQo-
hCG (W/L2 IS)
Fig. 4. Doubling times of serum hCG, samples taken 2-4 days to), and 5-8 days (+ ) apart at various stages of gestation. (*) represents DT values > 7 days, with number of occurrences next to it. The bars represent reference intervals established using log cumulative frequency.
288
Correlation Relationships between the assays were studied by linear regression analysis by arithmetic ratios. The slopes obtained with linear regression, forcing the through zero, were similar to arithmetic ratios for all three correlations (Table The ratio of values obtained with the two 2 IS assays, ABB to AMX, 1.39 + 0.28 (mean + SD) where the slope was 1.43, r = 0.960. Between the 1 assay (HYB) and the 2 IS assays, the ratios were 1.77 + 0.3 (AMX) and 1.31 f (ABB) while the corresponding slopes were 1.67 and 1.22.
and lines V). was IRP 0.29
Doubling time (DT) Doubling times of hCG levels taken two to four days (n = 68 pairs) and 5-8 days (n = 104 pairs) apart were calculated and grouped according to the second hCG level. As the distribution of DTs showed a log normal distribution, reference intervals were established using log probability paper as the 2.5 and 97.5 cumulative frequency values [13] (Fig. 4). Using the Student’s t test, there were no significant differences between the first three groups (hCG values between 0 and 17999 IU/l); for samples between 18 000 and 29999 IU/l the doubling times were significantly different from the first group. Discussion Reference curves established using our patients’ data showed the exponential increase seen previously by many investigators. However, the reference curves for the three assays differed from each other as well as from those supplied by the manufacturers. These differences did not arise from variations in technique as the mean values for the three levels of Lyphochek Human control sera from our laboratory were within one standard deviation of the group means from the QC manufacturer’s published summaries of other users. Of the three, the reference curve obtained for HYB was most similar to the manufacturer’s. With ABB, values were markedly different since the manufacturer’s values were not obtained using the assay itself. For AMX, there was discrepancy in the early part of the curve, which may be partly attributable to the manufacturer plotting upper and lower limit values at the beginning rather than the middle of each interval thus shifting the curve to the left. This difference is clinically significant, as up to 46% of samples from normal pregnancies taken between 28-35 days gestation could be interpreted as abnormally low using the manufacturer’s reference intervals. Shifting the curve to the right overcame this, but resulted in poorer detection rate of abnormal pregSince abortions are not usually life nancies, mainly in patients with abortions. threatening, this misclassification is generally not as significant as misdiagnosing a normal pregnancy as abnormal. Variations in reference values for different assays kits stem not only from the use of two WHO standards giving numerically dissimilar results but reported in the same units (IU/l), but from assay differences as well. The WHO 2 IS was prepared in 1963 for bioassays, and the 1 IRP or HCG 75/589 was prepared in 1975 for immunoassays. It is commonly accepted that values obtained with 1 IRP assays are 1.8 [14] to 2.07 [15] times that of 2 IS assay values. Lack of understanding of the
289
relationship between these WHO standards has led to confusion concerning the value of the ‘discriminatory zone’. This is exemplified by the fact that Kadar et al. [16] originally published their data for the “discriminatory zone” for serum hCG as 6000-6 500 IU/l 2nd IRP. This standard does not exist, but this value has subsequently been cited in the literature as applicable to both 2 IS and 1 IRP. Batzer et al. [7] in a prospective evaluation of 146 ultrasound observations on 98 pregnancies established 1000 ng/ml (or 5 600 IU/l 2 IS) as the upper limit. In a large prospective study, Kadar’s group [5] confirmed their previous findings, but stated that the hCG assay was calibrated against 1 IRP. Of 36 patients with ectopic pregnancies, 86% showed no sac with hCG levels above 6500 IU/l 1 IRP (or 3250 IU/l 2 IS). These authors erroneously concluded that their results were similar to those of Batzer, even though they were actually stating a value approximately half that of Batzer’s. In 1985, Nyberg [17], using a 2 IS hCG assay, reported a level of 1800 IU/l above which, the absence of an intrauterine sac was 100% sensitive for abnormal pregnancy. Unfortunately these studies have been widely quoted and misinterpreted with regard to standardization resulting in confusion and no clear consensus as to what the ‘discriminatory zone’ should be. Further complicating this issue is the considerable variability in sensitivity with ultrasound techniques. In addition to these problems of standardization, differences in hCG kits must also be considered. Variability in hCG values even when standardized similarly was reported as far back as 1983 when Rasor and co-workers [18] showed variable correlations between the National Institute of Health kit, three 1 IRP and seven 2 IS kits. They could not assign a relationship between the two WHO standards. The differences in values obtained using kits standardized similarly have been attributed to cross-reactivity with other gonadotrophins, and differences in hCG components detected. Commercial kits are now beta-specific to avoid cross reactivity with other pituitary glycoprotein hormones. However, considerable kit to kit variability may arise from the use of monoclonal versus polyclonal antisera as well as methodology. Of the three kits evaluated in this study, AMX uses a polyclonal antiserum which measures dimer hCG (intact (Y/P), free /?-subunits and fragments containing P-subunits. The ABB uses polyclonal antiserum, while HYB is monoclonal, and both measure only intact a/j3 dimer hCG. Hay [19] has shown that this variation in ligand specificity may yield discordant results when comparing assays. He postulated that discrepancies diminish with increasing gestational age. The immature trophoblast synthesizes predominantly free P-subunits; while mature cytotrophoblast produces sufficient a-subunits to match the P-subunits and yield mainly whole hCG dimer. There are also eight alleles for the P-subunit, so that even P-subunit specific assays may vary depending on whether monoclonal or polyclonal antiserum is used. This is often not considered when interpreting results involving serum hCG as long as kits are standardized against the same WHO standard. Recent papers comparing methods using the same standards, ABB versus AMX [20] and Stratus fluoroimmunoassay VS. HYB [21], have concluded that minor differences between the assays studied are analytically acceptable. Our study has shown that, with the kits evaluated, this may not be true for all kits. Values obtained with the ABB were on the average 1.4 times the value of
290
AMX. Both assays are standardized against 2 IS. The ratios between HYB which is standardized against 1 IRP and AMX and ABB were 1.8 and 1.3, respectively. The common practice of using published data generated by different assays [22-251 should be discouraged. A conversion factor may be used when a specific relationship between the two assays is known. However, caution should be exercised with results obtained using diluted samples, since matrix effects of dilution may make correlations nonlinear in the higher range. Using the rate of increase rather than absolute values may circumvent problems of assay variation and inaccurate gestational dates. However, considerable controversy has arisen as to the proper determination and interpretation of these values. Batzer et al. [9] obtained a DT of 2.2 + 1.0 days from the slope of the linear regression analysis of log hCG vs. gestational age during the first 44 days. Kadar et al. [16] determined the normal increase in serum hCG over 48 hours to be > 66%. In 1985, Pittaway et al. [lo] demonstrated that: (a) the exponential rate of increase in serum hCG decreases with gestational age and is more accurately represented by a quadratic equation, and (b) the DT of serum hCG increases with increasing hCG level or gestational age. They subsequently [26] showed that use of multiple values of DTs was more sensitive than use of a single mean DT determined by linear regression. These findings were confirmed by Daya [ll] who established values for three separate gestational periods, and Fritz [12] who proposed the use of a nomogram based on the level of hCG and the time between samples. This representation of hCG increase with gestational age has been criticized [27] as has the contention that DTs vary with gestational age, sampling interval and hCG. Kadar and Romero [28] have recently found that estimates of DT for serum hCG are not influenced by initial hCG, gestational age or sampling interval. Our data is consistent with a change in DT with gestational age; beyond 38 days, or 6000 IU/l 2 IS, DTs were unpredictable and of little use clinically. Furthermore, the use of DT is not always feasible. For patients with ectopic pregnancy, it may not be prudent to wait 48 h. In patients with abortion, samples taken more than 4 days previous may not be useful for determination of DT, as normal increases prior to the embryonal death may mask the fall off in hCG levels. In practice, we find evaluation of abnormal pregnancies with either doubling time or hCG value against reference interval alone inadequate in many patients. Plotting available values on a reference curve combines both methods, allowing us to evaluate each result against the reference interval as well as visually compare the observed rate of increase with the expected. Valid reference intervals are essential for this approach. Ideally each laboratory should establish its own reference intervals since even those determined by the manufacturer may not be appropriate. Unfortunately, with the rapid rise in levels in early pregnancy, a large number of patient samples is required to establish reliable reference intervals. It is imperative that clinicians be made aware of this variability in hCG assays and the pitfalls of using published data for interpretation of hCG values.
291
Acknowledgement The authors
thank
Amersham,
Hybritech
and Abbott
for supplying
kits.
References 1 Ooi DS, Perkins SL. Serum #IhCG assays in ectopic pregnancies (abstr.). Clin Chem 2 Cartwright PS, DiPietro DL. Ectopic pregnancy. Changes in serum human chorionic
1986;32:1166. gonadotropin
concen&ation., Obstet Gynecol 1984;63:76-80. 3 Braunstein GD, Asch RH. Predictive value analysis of measurements of human chorionic gonadotropin, pregnancy specific /3-l-glycoprotein, placental lactogen and cystine aminopeptidase for the diagnosis of ectopic pregnancy. Fertil Steril 1983;39:62-67. 4 Batzer FR, Corson SL. Diagnostic techniques used for ectopic pregnancy. J Reprod Med 1986;31:86-93. 5 Kadar N, Caldwell 6 7
8 9
10 11 12 13 14 15 16 17 18 19 20 21 22 23
BV, Romero
R. A method
of screening
for ectopic
pregnancy
and its indications.
Obstet Gynecol 1981;58:162-165. Romero R, Kadar N, Jeanty P, et al. Diagnosis of ectopic pregnancy: value of the discriminatory human chorionic gonadotropin zone. Obstet Gynecol 1985;66:357-360. Batzer FR, Weiner S, Corson SL, Schlaff S, Otis CV. Landmarks during the first forty-two days of gestation demonstrated by the /?-subunit of human chorionic gonadotropin and ultrasound. Am J Obstet Gynecol 1983;146:973-979. Nyberg DA, Filly RA, Filho DLD, Laing FC, Mahony BS. Abnormal pregnancy: early diagnosis by US and serum chorionic gonadotropin level. Radiology 1986;158:393-396. Batzer FR, Schlaff S, Goldfarb AF, Corson SL. Serial B-subunit human chorionic gonadotropin doubling time as a prognosticator of pregnancy outcome in an infertile population. Fertil Steril 1981;35:307-312. Pittaway DE, Reish RL, Colson Wentz A. Doubling times of human chorionic gonadotropin increase in early viable intrauterine pregnancies. Am J Obstet Gynecol 1985;152:299-302. Daya S. Human chorionic gonadotropin increase in normal early pregnancy. Am J Obstet Gynecol 1987;156:286-290. Fritz MA, Guo S. Doubling time of human chorionic gonadotropin (hCG) in early normal pregnancy: relationship to hCG concentration and gestational age. Feril Steril 1987;47:584-589. Neuman GJ. The determination of normal ranges from routine laboratory data. Clin Chem 1968;14:979-988. Rule AH, Michlewitz H, Boyle E, Donaboe M. Use of beta-human chorionic gonadotropin in gestational aging. Ann Clin Lab Sci 1985;15:428-434. Amerlex-M PHCG RIA kit insert. Amersham International plc, Amersham U.K. Kadar N, DeVore G, Romero R. Discriminatory hCG zone: its use in the sonographic evaluation for ectopic pregnancy. Obstet Gynecol 1981;58:156-161. Nyberg DA, Filly RA, Laing FC, Mack LA, Zarutskie PW. Ectopic pregnancy. Diagnosis by sonography correlated with quantitative HCG levels. J Ultrasound Med 1987;6:145-150. Rasor JL, Farber S, Braunstein GD. An evaluation of 10 kits for determination of human chorionic gonadotropin in serum. Clin Chem 1983;29:1828-1831. Hay DL. Discordant and variable production of human chorionic gonadotropin and its free (I- and P-subunits in early pregnancy. Clin Endocrinol Metab 1985;61:1195-1200. Hether NW, Gadsden RH. Evaluation of the Abbott p-hCG 15/15 enzyme immunoassay (EIA) kit with comparison to two other methods (Abst.). Clin Chem 1985;31:959. Beinlich CJ, Carpenter RH. Automated and manual quantitative assays of chorionic gonadotropin in serum compared. Clin Chem 1987;33:167-169. LeMaistre A, Bracey A, Katz A, Wu AHB. Role of qualitative chorio-gonadotropin assays in diagnosis of ectopic pregnancy. Clin Chem 1987;33:1908-1910. Hull ME, Schulman H. Impact of P-subunit human chorionic gonadotropin and ultrasonography in the management of suspected ectopic pregnancy. J Reprod Med 1986;31:478-482.
292 24 Pike Matthews C, Coulson PB, Wild RA. Serum progesterone levels as an aid in the diagnosis of ectopic pregnancy. Obstet Gynecol 1986;68:390-394. 25 Grudzinskas JG, Westergaard JG, Poulson H, Teisner B. The value of hCG measurements in the detection of ectopic pregnancy. Amersham International plc, Amersham U.K. 26 Pittaway DE, Colston Wentz A. Evaluation of early pregnancy by serial chorionic gonadotropin determinations: a comparison of methods by receiver operating characteristic curve analysis. Fertil Steril 1985;43:529-533. 27 Kadar N. Relationship between log of the human chorionic gonadotropin concentration and time period in early pregnancy [Letter]. Am J Obstet Gynecol 1986;154:692-693. 28 Kadar N, Romero R. Observations on the log human chorionic gonadotropin-time relationship in early pregnancy and its practical implications. Am J Obstet Gynecol 1987;157:73-78.