Establishing a reference interval for serum anti-dsDNA antibody: A large Chinese Han population-based multi-center study
Establishing a reference interval for serum anti-dsDNA antibody: A large Chinese Han population-based multi-center study
Chuiwen Deng 0 1 2
Shulan Zhang 0 1 2
Chaojun Hu 0 1 2
Ping Li 0 1 2
Ziyan Wu 0 1 2
Si Chen 0 1 2
Jing Li 0 1 2
Liubing Li 0 1 2
Fengchun Zhang 0 1 2
Yongzhe Li 0 1 2
0 Funding: This study was supported by the Research Special Fund for Public Welfare Industry of Health , No. 201202004
1 Editor: Xu-jie Zhou, Peking University First Hospital , CHINA
2 Department of Rheumatology and Clinical Immunology, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College , Beijing , China , 2 Key Laboratory of Rheumatology and Clinical Immunology, Ministry of Education , Beijing , China
Competing Interests: The authors have declared
that no competing interests exist.
This is the first exploration of the RI for anti-dsDNA antibody in the Chinese Han population.
We have established gender-specific RIs for each assay method commonly used in China.
In laboratory medicine, reference intervals (RIs) represent the typical fluctuations in the
quantity or quality of body fluid analytes in a relatively healthy population. The concept of an RI
was first proposed by Grasbeck et al. in 1968 [
], and it was initially called a ªnormal valueº.
Later, it was realized that the term ªnormalº was scientifically flawed. Then, well-defined
nomenclatures, including ªreference value,º ªreference range,º and ªnormal reference rangeº
came into use. From a statistical standpoint, the term ªreference intervalº better fits the
concept. Sometimes, an RI is confused with a clinical decision limit (CDL). A CDL is the threshold
concentration of a body fluid analyte, and a specific medical decision is made when the
concentration of an analyte for a given individual is above or below the CDL. Unlike an RI, a CDL
is obtained from clinical studies that explore the diagnosis or specific outcome of a certain
Generally, the manufacturers of diagnostic kits are obliged to provide the appropriate RI
for clinical laboratories. In diagnostic kits for autoantibodies, most manufacturers provide
cutoff values, which are used as RIs. However, not all RIs are rigorously calculated. One of the
major issues in the application of RIs has been the lack of standardization in the selection of
reference subjects. To address this problem, a standard protocol for establishing an RI
(C28-A3) has been proposed by the International Federation of Clinical Chemistry together
with the Clinical and Laboratory Standards Institute [
], and this has been widely used. In
addition, the RIs provided with kits are typically calculated using reference subjects from the
manufacturer's country or region, and they are not necessarily applicable to individuals in
other countries or regions. In China, most of the kits for autoantibody detection, which are
procured from outside China, do not provide RIs based on Chinese or Asian populations,
resulting in difficulties when evaluating RIs in clinical laboratories.
Fifty years ago, researchers found that circulating anti-dsDNA antibodies were present in
patients with systemic lupus erythematosus (SLE) [
]. Subsequently, anti-dsDNA antibodies
were shown to play important roles in SLE, both in its pathogenesis and as a biomarker for
diagnosis and prognosis [
]. Thus, anti-dsDNA antibodies were introduced as a diagnostic
biomarker in the classification and/or diagnostic criteria for SLE in 1982, 1997, and 2011 [
Then, a proposal was made that the criterion for the inclusion of anti-dsDNA antibody in the
classification of SLE should be modified. It was suggested that the anti-dsDNA antibody level
should be above the laboratory RI or twice the RI when tested by enzyme-linked
immunosorbent assay [
]. Thus, calculating an accurate RI for the anti-dsDNA antibody level is important
for making clinical decisions in SLE. Notably, there is a high incidence of SLE in China [
which makes it even more important to define an accurate RI for anti-dsDNA antibody in
To our knowledge, no study has explored the RI for anti-dsDNA antibody testing in a
Chinese population. In the present study, we aimed to recruit a large number of apparently healthy
Chinese Han individuals and establish RIs for anti-dsDNA antibody according to the
Selection of reference subjects
Since the characteristics of autoantibodies have been poorly studied, and the current literature
contains little relevant information, the factors that influence autoantibodies are little known.
Based on this background, we chose a posteriori sampling, which was recommended by
2 / 9
C28-A3 and fits the goal of our research. A posteriori sampling proceeds with the exclusion
and partitioning of participants after sampling and analyte testing.
Reference subjects were selected from a population using specific, well-defined criteria, as
recommended by C28-A3. In our study, we applied the following exclusion criteria: (1) a
diagnosis of autoimmune disease (SLE, etc.); (2) a diagnosis of a disease that can affect
immunoglobulin levels (liver cirrhosis, etc.); (3) activities or diseases that transiently affect the immune
system, such as excessive smoking (more than 20 cigarettes per day), having one alcoholic
drink every day for at least 2 weeks, surgery, or taking drugs that affect the immune system
(non-steroidal anti-inflammatory drugs, disease-modifying anti-rheumatic drugs,
immunosuppressors, glucocorticoids, gout suppressants, and biological agents), recent (prior few
months) blood transfusion or blood donation, pregnancy, obesity (body mass index 28),
confirmed malignancy, or diabetes. Since no RI for anti-nuclear antibody has been set, we
could not include the status of this autoantibody as a criterion for eligibility. Written informed
consent was not obtained for the essence of the study design. The utilized serum samples in
the research were taken from the leftover samples of the routine tests, which will not influence
the health or treatment of the participant. In this situation, verbal informed consent was
obtained from all subjects and this was recorded by the physician who explained the merits of
the project and the study procedure. The Ethics Committee of Peking Union Medical College
Hospital approved this study, including the consent procedure.
According to the standard protocol for establishing an RI (C28-A3), to derive a 95% RI
with the minimum subgroup of participants, 120 reference values, one from each reference
subject is required [
]. Crucial data were recorded, including age, sex, and ethnic background.
The information that authors had access to could not identify individual participants during
or after data collection. A physical examination and certain clinical laboratory tests were
performed as needed to assess the exclusion criteria and to confirm that the recruited participants
were healthy individuals.
From January 2012 to Jun 2014, apparently healthy adults of the Han population were
recruited from four hospitals in China, Peking Union Medical College Hospital, Guangdong
General Hospital, Shanghai Changzheng Hospital, and West China Hospital, representing the
Han populations from north, south, east, and west China, respectively. Finally, reference values
obtained from 2,880 apparently healthy adults were processed to calculate RIs for anti-dsDNA
Twelve hours prior to sample collection, participants were required to maintain their usual
daily habits and diet but to avoid drinking alcohol and smoking. Under a fasting condition in
the morning, blood samples were collected from the cubital vein and dispensed into 5-mL
pro-coagulation tubes with gel (Becton, Dickinson, and Company, Franklin Lakes, NJ, USA).
To separate the serum, blood samples were centrifuged at 3,000 rpm for 5 min within 6 h of
collection. Each serum sample was equally divided into 5 tubes and frozen at −80ÊC until use.
Assessment of the anti-dsDNA antibody tests
Analyte detection and statistical analysis were conducted from July 2014 to March 2016.
AntidsDNA antibody levels were tested with three kits that are commonly used in China and are
produced by EUROIMMUNE (Lubeck, Germany), AESKU (Wendelsheim, Germany), and
INOVA (San Diego, CA, USA). Briefly, serum samples were diluted in sample buffer and then
diluted serum, standards, and negative and positive controls were added to the designated
antigen-coated micro wells and incubated at room temperature. Between all incubation steps,
3 / 9
the wells were washed three times with wash buffer. Corresponding wells were probed by
adding the respective secondary antibody conjugates and incubated at room temperature. Then,
substrate was added to the respective wells and incubated at room temperature. Finally, stop
solution was added and the optical density of each well was read.
According to the C28-A3 guideline, a normal Q-Q plot was used to assess the data distribution
before the statistical analysis. Combining the consensus and Q-Q plot analysis data, the
measured anti-dsDNA antibody levels showed a skewed and non-Gaussian distribution. Therefore,
a nonparametric method was used to calculate the RIs as the lower and upper reference limits.
These two reference limits are the 2.5th and 97.5th percentiles of the distribution for the
reference population. In addition, we performed a bootstrap procedure with 1,000 replicates to
determine 90% CIs for both the lower and upper limits of the RIs.
The anti-dsDNA antibody RIs were calculated by the following two steps: First, a
preliminary RI was calculated using all the data and the outliers were removed. The Dixon method,
which is recommended by the C28-A3 guideline, has become widely used for identifying
outliers. The Dixon method involves the calculation of the ratio D/R, where D is the absolute
difference between an extreme observation (large or small) and the next largest (or smallest)
observation, and R is the range of all observations, including extremes. If the difference D is
equal to or greater than one-third of the range R, the extreme observation is considered an
outlier and should be deleted. After excluding outliers, supplementary participants were recruited
and anti-dsDNA antibody levels were measured. Second, the data were placed into subclasses
by gender, and the RIs for males and females were calculated. Then, a Wilcoxon rank sum test
was used to identify whether there were significant differences between the RIs. If there were
no significant differences (p > 0.05), the data from males and females were combined for the
statistical analysis. If there was a significant difference (p < 0.05), gender-specific RIs were
calculated. In addition, data were subclassified according to age (16±30, 31±40, 41±50, 51±60, 61±
70, and > 71 years) and district, and then the RI of each subclass was calculated.
The raw data of each group were utilized to perform comparisons between groups by
Wilcoxon rank sum test, and p < 0.05 indicates a significant difference.
The ages of the recruited apparently healthy adults ranged from 16 to 99 years. The study
sample was divided into groups by decade of age (i.e., 31±40, 41±50, and 51±60 years), except for
the groups of 16±30 and > 71 years old, which were set as the youngest and oldest age groups,
respectively. The average age of the study participants is shown in S1 Table. According to the
consensus view in the field [
], the minimum subgroup at one center should include 120
individuals (60 males and 60 females, S2 Table). After a posteriori sampling, 2,880 individuals were
enrolled from the four regions, with 1,440 males and 1,440 females. There were no significant
differences in mean age among the four regions or between the male and female groups.
RI of Chinese Han
The logarithms of all raw data to base 2 are shown as scatterplots (Fig 1). The gender-specific
RIs for anti-dsDNA antibody in the Chinese Han population are shown in Table 1. The
Wilcoxon rank sum test results for every tested assay kit showed that there was a significant
4 / 9
Fig 1. Scatter plots of the raw data for anti-dsDNA antibody. The logarithms of all raw data to base 2 were displayed on a scatter plot to visualize the
data distribution for each of the three tested assay kits. (A) distribution of the data from AESKU; (B) distribution of the data from EUROIMMUNE; (C)
distribution of the data from INOVA.
difference between the upper limits of the RIs for males and females (p < 0.05). Therefore,
gender-specific RIs were calculated.
RIs for different age groups and districts
There was an apparent age-dependent variation in the RI. Generally, the older age groups,
especially the group aged > 71 years, tended to have a higher upper limit of the RI (Table 2).
The levels of anti-dsDNA antibody were compared between males and females of the same age
group. Several age groups exhibited significantly higher levels of anti-dsDNA antibody in
females than in males; in particular, the age groups 31±40 and 41±50 showed this result for
each of the three tested assay kits.
The RIs for the four centers in China were calculated (Table 3). The levels of anti-dsDNA
antibody were compared between males and females at the same center. A few centers
exhibited a significantly higher level of anti-dsDNA antibody in females than in males, especially the
north and south centers, which showed this result for each of the three tested assay kits.
Comparison of the calculated RI and provided cut-off values
Among the three kits, the methods of defining cut-off value were diverse. The
EUROIMMUNE recruited 206 healthy blood donors and tested the levels of the anti-dsDNA antibodies.
They set the cut-off value as 100 IU/ml for 1.5% of the blood donors were anti-dsDNA
antibody positive when applying that value. In addition, INOVA declared that the normal range
* p<0.05, comparision of the levels of anti-dsDNA antibody between male and female in the same age group.
for their assay was determined by analyzing samples from 175 random blood donors. Of that
number all but 1.1% had anti-dsDNA antibody values less than 200 IU/ml. The AESKU did
not describe how the cut-off value was defined. All these cut-off values were not supposed to
be the upper limit of a RI (i.e., the 97.5% of a reference population). In our study, the
numerical values of the calculated gender-specific RIs and the cut-off values provided for the tested
assay kits were diverse. Most of the cut-off values were not suitable for clinical use. Many of
the 90% CIs that we calculated for the 97.5 percentile did not include the assay cut-off values,
although some of them did.
6 / 9
* p<0.05, comparision of the levels of anti-dsDNA antibody between male and female in the same center.
Identification and handling of outliers
According to the Dixon method [
], there were four outliers in the AESKU assay, and one in
the EUROIMMUNE assay [
]. No outliers were observed in the INOVA assay. Therefore,
five additional apparently healthy adults were added to replace the outliers (Table 4).
The RI is important for clinical decision-making, including diagnosis, evaluation of disease
activity, and prognosis. Clinical decision-making for autoimmune diseases mainly relies on
the detection of autoantibodies. Typically, the manufacturer of a test kit provides cut-off
values, which are used as the RI in clinical practice. However, most of the kits currently used to
measure autoantibodies in China are imported kits, and their cut-off values have been derived
from foreign populations and have not been systematically calculated. Differences in ethnic
background or race could lead to a disparity in the RIs for immunologic indices, for example,
serum autoantibodies [
]. Although anti-dsDNA antibody is an important immunologic
index for autoimmune disease, its RI has not been systematically defined until now. Here, we
conducted this study to establish RIs for anti-dsDNA antibody in the Chinese Han population
according to guideline C28-A3.
In this research, gender-specific RIs for anti-dsDNA antibody were observed in the Chinese
Han population (Table 1). Generally, cut-off values are set at a point that is convenient for
clinical use. In our study, the RIs for anti-dsDNA antibody were far higher or lower than the
cutoff values provided by the manufacturers of the tested assay kits. Many of the 90% CIs that we
calculated for the 97.5 percentile did not include the assay cut-off values, although some of
them did. The differences between the provided cut-off values and our calculated RIs may be
explained by two factors. First, the cut-off values were calculated using non-Chinese
populations. Second, the methods for calculating cut-off values, including recruiting participants and
statistical analysis, have not been standardized among kit manufacturers. Our study
established accurate and detailed RIs for anti-dsDNA antibody testing according to the standard
protocol, C28-A3, and these RIs could help clinicians to increase the accuracy of their clinical
It was interesting to find that, in the age groups of 31±40 and 41±50 years, the levels of
antidsDNA antibody were higher in females than in males (Table 2). In addition, significantly
higher levels of anti-dsDNA antibody were found in females recruited from the north and
south centers than in males from the same centers (Table 3). These results suggest that
unknown factors may influence anti-dsDNA antibody concentrations in the above-mentioned
age groups or centers, and specific cut-off values should be set for them. In addition, we
observed that the older age groups tended to have higher values of RIs, particularly the
population aged 71 years and older. We speculate that changes occur with age in females that result
in abnormal immune function, especially the production and clearance of autoantibodies.
Of note, the significant differences that we discussed above might not only be attributed to
gender, age, and region, but could also result from random variance and others. Therefore,
additional related studies should be performed to explore the pathogenesis of autoimmune
disease and provide more information regarding other factors that influence the RI for
There were five outliers (Table 4), which were replaced with new participants. Since we
ruled out participants with autoimmune diseases by a physical examination and clinical
laboratory tests, the anti-dsDNA antibody levels of these outliers might be elevated for other
physiological reasons, or the elevation might be indicative of the onset of related autoimmune
diseases, as other autoantibodies, such as anti-Sm antibody, can be detected several years
before disease onset. On the latter basis, these five outliers warrant long-term follow up.
In conclusion, this is the first study to have explored the RI for anti-dsDNA antibody in the
Chinese Han population. Gender-specific RIs were established, and their application may help
clinicians to make accurate decisions for Chinese Han individuals.
S1 Table. Average age of recruited participant from different centers of China.
8 / 9
S2 Table. The number of participant recruited from different centers of China.
Conceptualization: FCZ YZL.
Data curation: CWD YZL.
Formal analysis: CWD.
Funding acquisition: YZL FCZ.
Investigation: CWD SLZ CJH PL ZYW SC JL LBL.
Project administration: YZL FCZ.
Software: CWD SLZ.
Validation: CWD CJH.
Writing ± original draft: CWD.
Writing ± review & editing: CWD FCZ YZL.
1. GraÈsbeck R , Fellman J . Normal values and statistics . Scand J Clin Lab Invest . 1968 ; 21 : 193 ± 195 . PMID: 5708691
2. Sikaris KA . Physiology and its importance for reference intervals . Clin Biochem Rev . 2014 ; 35 :3± 14 . PMID: 24659833
3. Horowitz G , Altaie 1S , Boyd JC , Ceriotti F , Garg U , Horn P , et al. Defining, establishing, and verifying reference intervals in the clinical laboratory . Approved guideline , 3rd ed. Clinical Laboratory and Standards Institute; 2008 .
4. Holman HR , Kunkel HG . Affinity between the lupus erythematosus serum factor and cell nuclei and nucleoprotein . Science . 1957 ; 126 : 162 ± 163 . PMID: 13442661
5. Isenberg DA , Manson JJ , Ehrenstein MR , Rahman A . Fifty years of anti-ds DNA antibodies: are we approaching journey's end? Rheumatology (Oxford) . 2007 ; 46 : 1052 ± 1056 .
6. Fu SM , Dai C , Zhao Z , Gaskin F . Anti-dsDNA Antibodies are one of the many autoantibodies in systemic lupus erythematosus . F1000Res . 2015 ; 4 : 939 . doi: 10.12688/f1000research.6875.1 PMID: 26594353
7. Petri M , Orbai AM , AlarcoÂn GS , Gordon C , Merrill JT , Fortin PR , et al. Derivation and validation of the Systemic Lupus International Collaborating Clinics classification criteria for systemic lupus erythematosus . Arthritis Rheum . 2012 ; 64 : 2677 ± 2686 . doi: 10 .1002/art.34473 PMID: 22553077
8. Zou YF , Feng CC , Zhu JM , Tao JH , Chen GM , et al. Prevalence of systemic lupus erythematosus and risk factors in rural areas of Anhui Province . Rheumatol Int . 2014 ; 34 : 347 ± 356 . doi: 10 .1007/s00296- 013 -2902-1 PMID: 24264010
9. Mok CC , To CH , Ho LY , Yu KL . Incidence and mortality of systemic lupus erythematosus in a southern Chinese population, 2000 ± 2006 . J Rheumatol . 2008 ; 35 : 1978 ± 1982 . PMID: 18688913
10. Dixon WJ . Processing data for outliers . Biometrics . 1953 ; 9 : 74 ± 89 .
11. Schauer U , Stemberg F , Rieger CH , Borte M , Schubert S , Riedel F , et al. IgG subclass concentrations in certified reference material 470 and reference values for children and adults determined with the binding site reagents . Clin Chem . 2003 ; 49 : 1924 ± 1929 . PMID: 14578325