A-SAAE-Selectin

ICAM1

IFNgIL1ra

IL2RIL6 IL7MMP1

MMP3

MMP9RANTES

TIMP1

TIMP2TNFRI

TNFRII

MD

:RD

Ra

tio 1

2

5

10

20

30

RF Lowest Tertile

RF Middle Tertile

RF Highest Tertile

hMMP1

1 2 3 4 5 6 7 8 9

MD

:RD

Ra

tio

0.5

1

2

5

10

20

30

50

RF

(IU)

0

200

400

600

800

Rheumatoid Factor Interferes with Multiplex-Based ELISAIn Patients with Rheumatoid Arthritis

Derrick J. Todd1, Nicholas Knowlton2, Elizabeth W. Karlson1, Nancy A. Shadick1, Michael E. Weinblatt1, Peter H. Schur1

Ronenn Roubenoff3, Elena Izmailova4, Michael Centola2, David M. Lee1

1Division of Rheumatology, Immunology, and Allergy, Brigham and Women’s Hospital, Boston, MA; 2Oklahoma Medical Research Foundation, Oklahoma City, OK3Biogen-Idec Pharmaceutical, Cambridge, MA; 4Millennium Pharmaceuticals, Cambridge, MA

Introduction

Heterophilic antibodies are known to interfere with someimmunoassays. Agents such as Heteroblock® have beendesigned to suppress non-specific signals from heterophilicantibodies. Rheumatoid factor (RF) is a heterophilic antibody thatis present in the serum of up to 80% of patients with rheumatoidarthritis (RA). Multiplex-based enzyme-linked immunosorbantassays (ELISA) allow serum to be tested for multiple analytes.This emerging high-throughput technique is used with increasingfrequency in the study of serum from patients with RA.

Objective

We sought to answer whether multiplex ELISA arrays areaffected by serum RF or other heterophilic antibodies in patientswith RA.

Methods

Multiplex ELISA of RF-depleted Serum Samples•Serum was obtained from 9 RA patients. Aliquots of eachsample were then RF-depleted (RD) or mock-depleted (MD) byaffinity absorption against human IgG-conjugated orunconjugated sepharose, respectively. All absorptions wereincubated vol:vol >4 hr at 4°C. RF levels were measured bynephelometry.

•Using SearchlightTM multiplex ELISA, RD and MD samples weretested in duplicate for 16 analytes: A-SAA, E-selectin, ICAM1,IFNγ, IL1ra, IL2R, IL6, IL7, MMP1, MMP3, MMP9, RANTES,TIMP1, TIMP2, TNFRI, and TNFRII (PerBio/ThermoFisher,Woburn, MA).

•RF interference (RFI) was defined as ≥2-fold false elevation ofanalyte signal in the non-RF-depleted sample (MD:RD ≥2).

Heteroblock® and Cross-bead Multiplex ELISA•Serum was obtained from 3 RA patients and incubated for 2 hrat 25°C with 0, 1.5, 15, 150, or 1500 µg/ml Heteroblock® (OmegaBiologicals, Inc., Bozeman, MT).

•Using Multiplex Bead-based Luminex® Assays, serum sampleswere tested in duplicate for 23 analytes: Eotaxin, GM-CSF, IFNα,IFNγ, IL1β, IL1ra, IL2, IL4, IL5, IL6, IL7, IL8, IL10, IL12, IL13,IL15, IL17, IP10, MCP1, MIP1α, MIP1β, RANTES, and TNFα(Invitrogen, Carlsbad, CA). To test non-specific analyte-independent signal, cross-bead assays were also performedusing analyte-inappropriate beads.

•Multiplex Bead-based Luminex® Assays were also used to testserum samples from an additional 35 RA patients for the same 23analytes. Samples were incubated for 2 hr at 25°C with 0 or 150µg/ml Heteroblock®. These assays also included a positivecontrol that was “spiked” with a known quantity of analyte.

Results

•Following RF depletion, analyte signals were often reduced. Inaddition, the degree of signal reduction generally correlated withbasal RF titer, as shown for MMP1 (Figure 1, top panel). Forpatients in the lowest, middle, and highest basal RF tertiles, themean MD:RD ratio was ≥2 in 2/16 (12.5%), 4/16 (25%), and 15/16(93.8%) analytes, respectively (Figure 1, bottom panel).

•Non-specific signals for MIP1α were detected in 2/3 RA patientswhen assayed in cross-bead analysis (Figure 2). This signal wasdampened by high concentrations of Heteroblock®. The other RApatient demonstrated little if any non-specific signal.

•In a substantial number of the 35 RA patient samples tested byMultiplex Bead-based Luminex® Assays, analyte signals weredampened by Heteroblock® (represented by MIP1α, TNFα, andIL17 in Figure 3). Most analytes showed interference, but it wasnot uniformly present (represented by IL8 in Figure 3).

Conclusions

•In RA patients, serum RF and heterophilic antibodies interferewith analyte measurements in multiplex-based ELISA arrays.

•Large concentrations of Heteroblock® are needed to dampennon-specific interfering signals.

•These findings should be considered when interpreting datagenerated by multiplex arrays in patients with RA.

AcknowledgementsFunding support: NIH grant T32 AR007530, ACR REF Fellowship Training Award,Biogen-Idec, Millennium Pharmaceuticals, and Crescendo Bioscience

MIP1!

Heteroblock (ug/ml)

0 1.5 15 150 1500

Mean F

luore

scence In

tensity

0

1000

2000

3000

4000

350, Specific

262, Specific

292, Specific

350, Non-specific

262, Non-specific

292, Non-specific

MIP1!

Unblocked (MFI)

100 1000 10000

Blo

cked (M

FI)

0

200

400

600

800TNF!

Unblocked (MFI)

1 10 100 1000 10000

Blo

cked (M

FI)

0

50

100

150

200

250

IL-17

Unblocked (MFI)

0.1 1 10 100 1000 10000 100000

Blo

cked (M

FI)

0

5

10

15

20

25IL-8

Unblocked (MFI)

10 100 1000

Blo

cked (M

FI)

0

100

200

300

400

500

600

Figure 3. Heteroblock Reduces Non-specific Signal.Shown are 4 representative analytes tested by Multiplex Bead-based Luminex®

Assays in 35 RA patients (cyan circles) and one “spiked” positive control with aknown concentration of analyte (magenta circle). Samples were incubated withHeteroblock (y-axis) or not (x-axis). A reference line is included (y=x). Circles to theright of the line indicate a sample with signal that was dampened by Heteroblock®,sometimes markedly so.

Figure 1. Elevated basal serum RF is associated with interference.Top Panel: Shown in blue is the ratio of MMP1 signal in mock-depleted (MD) andRF-depleted (RD) serum samples from 9 RA patients. Also shown are RF values(solid line) and the defined threshold for RF interference (dashed line, MD:RD = 2).Bottom Panel: Shown are MD:RD ratios for all 16 analytes. For analysis, patientswere grouped into tertiles based on serum RF level.

Patient ID

Figure 2. Non-specific Signal by Cross-bead Analysis.Serum samples from 3 RA patients were incubated with various concentrations ofHeteroblock® (x-axis). Shown are MIP1α signals as analyzed by Multiplex Bead-based Luminex® Assays using analyte-specific beads (green) and non-specificcross-beads (red). Results for MIP1α were representative of other analytes.