A guide to the systematic approach needed to overcome false signals and optimize test performance.
In recent years the IVD industry has increased enormously its work to develop membrane-based lateral-flow tests. Such tests have found applications in both clinical and nonclinical fields. A review of the wide range of applications for these devices has been reported in an earlier article.1
While the concept of membrane-based lateral-flow immunoassay tests is quite simple, how well such tests perform depends on a number of critical parameters. Earlier articles in IVD Technology have discussed the importance of the quality of the gold conjugate and the way that capture proteins bind to membranes.1– 3
This article explores the possible causes of poor test performance that result in the generation of false-positive and false-negative signals, and provides a systematic approach to overcoming such problems and optimizing test performance. While this article specifically describes such problems with specific reference to gold-based lateral-flow tests, very similar problems are also found in latex-based tests.
In order to avoid repetition, it is assumed that the reader is already familiar with the mechanics of a lateral-flow test and with the basic properties of the lateral-flow test components—specifically the antibodies, gold conjugate, membrane, conjugate release pad, and sample and absorbing pads—as well as the standard chemical components that are used to treat the various papers and membranes used in the test. For the reader unfamiliar with these concepts, they have been described in detail elsewhere.1
Symptoms of False-Positive and False-Negative Results
A false-positive result is the appearance of a colored line on the test strip (red for gold conjugates) in the absence of analyte in the sample. The line may appear early in the test or after a significant time delay. A false-negative result is the absence of a visible line at the capture antibody (for a sandwich assay) in the presence of a detectable positive sample.
Either type of false signal may occur with a wide variety of samples, or may arise only with certain types or sources of sample. The problem may appear in every test strip from a single batch of devices, or it may only occur in a random number of strips within a sample. The latter situation demonstrates the importance of performing large trials of test strips at each stage of development before moving on to the next stage toward full manufacture.
Even at the level of full manufacture, it may be found that occasional false-positive results are obtained from standard negative samples and vice versa. Because of the enormous cost of batch failure, it is therefore mandatory to test thoroughly for, and understand the causes of, false-positive and false-negative results well before reaching large-volume production.
False-positive and false-negative signals may have many different causes. Such signals may be caused by either the sample or by the poor or inadequate design and manufacture of the test itself. The potential for generating false signals should be eliminated during the course of a test's development through a thorough, iterative trial of large numbers of test devices at each stage of development.
The cause of a false-positive or false-negative result can sometimes be determined from the visual appearance of the signal, or of the flow characteristics leading to the signal, as the test is performed. However, the reason a false signal has arisen will not always be evident. Only experience, a systematic and scientific approach, and a detailed troubleshooting schedule can eliminate such signals and prevent them from happening. Before adjusting any particular parameter, the test developer should be thoroughly familiar with what makes a test work reliably and what can cause it to go wrong. It is not necessary to adopt an empirical approach if a full understanding of the mechanics of a lateral-flow test has been achieved.
A detailed description of the workings of a membrane-based lateral-flow test and a detailed troubleshooting manual for false-positive and false-negative results is outside the scope of this short article.4 Instead, this article summarizes typical symptoms and describes a methodological approach for determining the cause of the problem. Suggested causes and cures are also provided.
Basic Causes of False Positives
Most causes of false-positive signals arise for the very same reasons that proteins bind specifically to gold particles during a routine conjugation procedure. During such a conjugation, proteins are adsorbed onto the surface of the gold by three major forces (see Figure 1).
|Figure 1. Binding forces between an anitbody and a gold particle.|
Charge Attraction. Gold particles are negatively charged because of a layer of negative ions adsorbed onto their surfaces from the reducer (often citrate) used to convert the gold salt into gold colloid. This negative charge will attract positively charged proteins and bring them close enough to the surface for the binding forces to take effect. Proteins that are more acidic than their isoionic point will be positively charged and are thus likely to be strongly attracted to the gold surface. Regions of the protein particularly rich in lysine or arginine will be strongly positively charged at pH values lower than the pH of lysine (pH 10.4) and arginine (pH 12.5).
Hydrophobic Binding. Once the proteins are close enough, (that is, closer than approximately 1 nm), any hydrophobic areas of the protein will be more likely to make contact with and bind to the hydrophobic gold surface. Any protein rich in apolar amino acids (e.g., tryptophan, valine, leucine, isoleucine, or phenylalanine) will thus be strongly bound to the gold surface.
Dative Bonding. Dative bonding provides the strongest binding of all. Proteins with a high content of sulphur-containing amino acid groups (due to cysteine and methionine residues) will bind strongly to the surface of gold. This is because of the attraction between gold atoms (having conductive electrons) and sulphur atoms (having valence electrons).
While these three types of binding forces work in favor of the production of stable gold conjugates, they also work against the performance of the assay system by creating false-positive signals as described below.
Before adopting a systematic approach to diagnosing and curing false-positive signals, it is important to be aware of most, if not all, of the possible reasons for such signals. Figure 2 schematically illustrates the major areas of concern, while the descriptions below summarize the most common causes. However, these sources are not at all exhaustive. Generally, all possible causes of false signals will be connected with either the gold conjugate, the capture antibody, the membrane, the added chemicals, or the sample. By understanding these potential causes, a speedy diagnosis can be made from the symptoms that arise during the performance of the test.
|Figure 2. Main sources of false positives.|
Problems Arising from the Gold Conjugate. Areas of the gold particles may become partly uncovered by proteins for any of several reasons. Antibodies may be removed while drying the gold conjugate or during the test procedure, or the conjugate may have been poorly coated in the first place. For any of these reasons, naked gold regions will be attracted to any positively charged protein or nitrocellulose or nylon membrane. This will be especially true as the gold passes through the capture line, since distances are very small and the likelihood of physical contact is great. It is therefore very important to coat the gold particles effectively and thoroughly so that the proteins do not detach during storage, handling, or the performance of the test.
Apart from the attraction of gold to the capture line, the conjugated antibody itself may be attracted to the capture line under certain operating conditions. This may again be due to charge or hydrophobic interactions, and may depend on the acidic or ionic environment during the procedure.
Excess gold conjugate will create several problems. Primarily it will increase the possibility of false-positive signals at the capture line simply because of the large quantity of gold conjugate passing the line. It will also greatly increase the likelihood of backflow of gold after the test period has elapsed. A good-quality gold conjugate should not need to be used in excess.
Gold conjugate might also be clustered, for a variety of reasons, usually through poor manufacturing. Large-enough clusters may block the membrane at any point where there is a restriction. If clustering is caused by the gold being hydrophobic, then the gold particles may also stick to the capture antibody during the flow of the conjugate.
The gold-labeled antibody may react nonspecifically and immunologically with the capture antibody regardless of whether analyte is present. While such reactions between a pair of antibodies may not always be detected in ELISA techniques, the method of forcing labeled antibody to flow very close to capture antibody in a nitrocellulose membrane increases this possibility. In addition, especially where polyclonal antibodies are used, there may be some nonspecific reactivity with other analytes in the sample, depending on the purity of the antibodies.
False-Positive Signals Associated with the Capture Antibody. A variety of factors cause some capture antibodies to behave in a nonspecific sticky manner. Such behavior may be due to hydrophobicity, nonspecific immune reactions, additives within the antibody, high positive charge, or a high concentration of sulphur-containing amino acids in the capture protein which will attract the gold. The same forces that cause antibodies to adsorb and bind onto the surfaces of gold particles during conjugation can actually work here with the capture antibody to the detriment of the lateral-flow test performance.
Difficulties Associated with the Solid-Phase Materials. Nitrocellulose membranes are extremely fragile and are easily damaged by contact. It is therefore very important to avoid any mechanical contact with the membrane surface during the procedure for striping the capture antibody. If the membrane is compressed at the capture line while the capture antibody is being applied, there is an increased risk of nonspecific trapping of gold conjugate at that point during the flow of the sample.
Two things can cause residual gold to stick at the capture line—too slow a flow of the gold along the membrane, or too slow a release from the conjugate pad. These events can happen if the membrane has too small a pore size, if there is not sufficient surfactant in the strip, or if there is poor contact between the membrane and the gold conjugate pad or the absorbing pad. Also, the membrane may be hydrophobic, thus hindering smooth flow of the gold conjugate. In addition, some samples can be very viscous (serum, for instance), which also slows down the flow. Gold can also stick at the capture line when there is an insufficient amount of sample or of surfactant in the system to wash all the gold along the strip.
If the test takes too long to read (usually anything longer than 15 minutes), there is a possibility that excess gold conjugate will start flowing back down the strip from the absorbing pad onto the membrane as the latter dries out. Gold returning from the absorbing pad is very likely to dry out at the capture line because during drying the capture antibody becomes very hydrophobic.
Risks Associated with Added Chemicals. Some lateral-flow test manufacturers perform blocking of membranes as a matter of course. With certain membranes, and when applying certain types of sample, blocking procedures convert the membrane from a chromatographic separator (bibulous) to a nonchromatographic strip (nonbibulous). This conversion is designed to improve the flow of the sample and the gold conjugate along the membrane strip.
However, blocking a membrane by immersion in a protein or surfactant solution is also likely to wash out any additives that the manufacturer incorporated to prevent the membrane from drying completely and becoming hydrophobic. Such blocking may thus cause the membrane to behave in a more hydrophobic manner when dry and can even cause more-general background staining or false-positive signals at the capture line.
Alternatively, blocking with excess protein or surfactant can produce high viscosity during sample flow and may reduce the sample clearance rate. Blocking with the wrong kind of reagents can also change the characteristics of the capture antibody, making it more sticky through charge, hydrophobicity, or increased sulfur-hydrogen (SH) attachments. Blocking should only be used for demonstrably good reasons, such as when the flow of sample or gold needs improving, and then only with minimum reagent concentrations.
Some preservatives, whether in the capture antibody, labeled antibody, or sample, can produce false positives. Thimerosal, which contains both sulphur and mercury, and lysine, which is always very positively charged at pH <10.4, are particularly troublesome.
Problems Specific to the Sample. Many samples contain components which may bind nonspecifically to the gold conjugate or to the capture line, and in doing so may produce nonspecific results. For example, samples may contain bacteria that may be partly broken down into cellular fragments and that may be extremely hydrophobic. These hydrophobic cellular fragments may also produce cross-linking between the capture line and the gold conjugate.
Other samples may contain high levels of sulphur or SH groups, or be very positively charged. In addition, some samples may contain large enough molecular or cellular components to block the membrane and disturb the flow of gold along the strip.
Samples may vary greatly in their acidity. For example, urine samples may be initially presented at pH 4–7 and may, in the absence of preservative, gradually become more acidic as bacterial contamination increases. Acidic samples will produce a positive charge on the capture antibody during sample flow and, in turn, this will cause the nonspecific attraction of negatively charged gold conjugate.
If samples contain large quantities of acidic or positively charged proteins, they may bind nonspecifically to the gold conjugate before ever reaching the capture line. This can either mask the conjugate, thus reducing the specific signal, or enlarge the conjugate complex to such an extent that clustering occurs on the membrane and at the capture antibody.
Diagnosis and Remedies for False Positives
When faced with either occasional or recurrent false-positive signals, a systematic approach to diagnosis of the problem should be adopted in order to apply the most appropriate remedy (see Figure 3). The most obvious causes should be looked at first. These are problems with reactions between the gold conjugate and the capture antibody; specifically charge attraction, hydrophobicity, and gold-sulphur bonding.
Next, nonspecific cross-reactivity between antibodies, specific sample characteristics, and membrane flow characteristics should be observed. The appropriate use of controls will quickly provide a guide to the source of the problem.
Following is a description of a typical systematic method of diagnosis, together with reference to the possible causes listed previously. As with any method of systematic diagnosis, only one parameter should be changed at a time when performing controls. In this way, a process of elimination can take place. A suggested sequence of questions follows.
1. Is the problem due to charge? Varying the acid levels in the test system (from pH 5 to 11) will demonstrate whether positive charge is occurring somewhere and attracting gold to the capture line.
2. Is the problem due to hydrophobicity? This may occur in the solid phase, the capture line, or the gold conjugate. Varying surfactant concentrations in the system will give clues to whether hydrophobicity is the cause.
3. Is the problem gold-SH attraction? This is most likely to occur from cysteine and arginine groups within the capture line or the sample. Closer inspection of these two regions as described in the following section will reveal the difficulty.
Systematic Approach to Problem Solving for False Positives
For most of these possible scenarios, systematic testing can be performed using only a dipstick (or "half stick"). The fully dried test strip is not necessary. To conduct such tests, the nitrocellulose membrane strip is placed directly into a microwell containing the gold conjugate, the sample, and the chemicals that would normally be dried into the fully assembled device. In this way, several tests may be made quickly and easily without the need for full assembly and drying. However, if the problem arises from the drying procedure, then the full assembly must be performed before testing.
The questions to be asked from the systematic approach can be grouped into the following five categories associated with the test assembly and components.
Is the Problem Related to the Gold Conjugate? This can be easily determined by using an alternative gold conjugate—for example BSA-gold at the same concentration and at the same acidity level as the original conjugate. If the problem persists, it is most likely to be caused by a charge effect. If the false signal disappears with an alternative gold conjugate, then it was most likely caused by a poorly made original conjugate or a problem with a conjugated antibody. Other controls to use here would be similar-species conjugates and alternative-species conjugates of monoclonal and polyclonal antibodies. The test developer should always have a range of alternative gold conjugates available for such testing.
An example of detecting problems stemming from the gold conjugate follows. A test was developed for an infectious disease using clinical samples (serum). False-positive results were observed for all negative samples. The signals disappeared when using a BSA-gold conjugate. With alternative nonspecific conjugates, the signals also disappeared. When changing to nonclinical control samples, however, such as PBS at pH 7.2, the problem persisted with the specific gold conjugate but not with the others. Changing the acidity of the buffer to pH 10 reduced the occurrences of false positives, but did not cause them to disappear. It was thus concluded that the problem was independent of the sample or capture antibody and was due instead to the high sensitivity of the gold conjugate specific to the capture antibody. The gold conjugate was suspected of having naked gold areas which bound to the capture line. Using a freshly and carefully made gold conjugate eliminated the false-positive signals.
Is the Problem in the Capture Antibody? Suitable controls to use in these situations would be alternative capture antibodies from similar or alternative species. Before using such alternatives, however, striping a capture protein of BSA alone will demonstrate whether the problem lies elsewhere. It may also be possible that it is not the antibody itself that is causing the false signal, but any preservative that may be in the antibody. In this case, dialysis against a suitable buffer (e.g., 10 mmol PO4) can improve matters. If the problem disappears with alternative antibodies, then clearly the specific capture antibody is the cause of the false signals and must either be replaced or cleaned up. For example, in practice, polyclonal antibodies from rabbits are sometimes particularly troublesome because of their hydrophobicity and should be treated with extra care.
A pregnancy test for detecting ßhCG in urine was found to produce false-positive signals for all samples, regardless of whether they were positive or negative. Altering the acidity levels did not reduce the signal, neither did changing the gold conjugate. The false-positive signals persisted even with a PBS buffer control sample.
It was suspected that the problem lay in the capture antibody. Striping the capture line with BSA instead eliminated all signals. Dialysis of the original capture antibody against a 10 mmol PO4 buffer eventually produced signals faithful to the sample. It was concluded that the capture antibody had been suspended in a buffer containing SH components such as thimerosal, and that these need to be removed or avoided.
Is the Problem Related to the Membrane? A significant part of any test development process is the correct choice of membrane. Some membranes are more suited to certain types of assay or samples than others. The developer should have a full repertoire of membranes from all manufacturers and in a range of pore sizes for immediate comparison. For example, random false-positive signals have been observed from test strips cut from batches of membrane where macroscopic hydrophobicity has occurred during drying.
The choice of membrane for a lateral-flow test should consider not only the desired flow rate and protein-binding characteristics, but also the homogeneity of the membrane during the preparation processes. Suitable controls to use here would be membranes of different pore sizes, different sources, and different areas of the same membrane batch.
For a serological test designed to detect antibodies to H. pylori, initial test development up to experimental prototype stage produced no false-positive or false-negative results. Full field trials determined that the test was ready to move on to the production prototype stage.