ZAP-70. The name alone is zesty enough to inspire
hope. Here, at last, is a protein that could serve as a pivotal prognostic
marker for patients with chronic lymphocytic leukemia.
As it turns out, however, the only thing zippy about ZAP-70 is its name. Otherwise, this tyrosine kinase has created an endless series of headaches for clinical labs as they try to extrapolate research data from expression arrays to more manageable protein expression assays.
Here’s the short version of why it’s so hard. ZAP-70 is labile. It’s an intracellular
antigen. It’s dim beyond belief. There’s no gold-standard assay. No one knows
for sure how to report it. Flow cytometry versus immunohistochemistry?—It
might be a tossup, no matter what people say. The antibodies leave much to be
desired. And clinicians who are lucky enough to get a result from their labs
probably don’t know what to do with it.
And here’s the short version of why ZAP-70 will nevertheless stick around, at least for the foreseeable future: It looks like it could replace immunoglobulin heavy-chain gene mutation status as a means for deciding which CLL patients need to be treated sooner rather than later.
Now for the long version.
ZAP-70 has inspired nonstop questions, discussions, and debates, most recently at the Clinical Cytometry Society meeting in Long Beach, Calif. Few concrete answers emerged, but not for lack of trying. In fact, the regularly scheduled programs were overshadowed by an impromptu meeting convened at the behest of Gerald Marti, MD, PhD, who had spent no small portion of his time perusing, "rather critically," he says, the posters related to ZAP-70.
Of the half dozen he viewed, two were of little immediate consequence, he says,
because they provided no detail on ZAP-70 determinations. The remaining four
did and were compelling enough to prompt Dr. Marti to ask the meeting’s chair
to find an empty conference room and to let attendees know that on Monday evening,
the 5:30-7:00 block, normally tantamount to dead airtime, would be devoted to
an ad hoc focused discussion on ZAP-70.
Though crowd estimates vary, all agreed the room was filled, an impressive feat for an off-the-cuff, word-of-mouth event. "It confirms how big the interest is in this problem," says Dr. Marti, who bears an astonishingly long title: chief, flow and image cytometry section, laboratory of stem cell biology, Division of Cell and Gene Therapies, Office of Cellular, Tissues and Gene Therapies, Center for Biologics Evaluation and Research, Food and Drug Administration.
Dr. Marti invited one speaker from each of the four posters to talk about their work; those in the audience weighed in with questions and observations. Dr. Marti, for his part, manned the flip chart, listing issues and problems lobbed by the crowd. "The list grew with each succeeding minute," recalls one attendee.
Scientific progress, like a good novel, is rarely transparent.
At the meeting, the air was filled with advice, giving fresh hope to those grappling with the assay. Anand Lagoo, MD, PhD, director of the flow cytometry laboratory and assistant professor of pathology, Duke University Medical Center, Durham, NC, counts himself among those who ferried pointers back to their institutions hoping for a breakthrough. But so far all have fallen short. "We have already started doing some of the things they were suggesting," said Dr. Lagoo in a phone interview two weeks after the meeting. "And everything they said will work is not working, so we’re still struggling with it."
Vincent Shankey, PhD, was also in the audience. Dr. Shankey, a senior staff development scientist at Beckman Coulter’s Advanced Technology Center, Miami, recalls that "a lot of the issues were very nicely discussed" at the meeting—but, as Dr. Lagoo’s experience bears out, none of it led to definitive answers.
At the end of the meeting, the only tangible conclusion was an agreement to study the matter further. Usually that’s a death rattle, like politicians creating a "panel of experts" to "further assess the matter," which really means constituents will never find out where their tax dollars disappeared to.
But in the case of ZAP-70, no one’s giving up, even if immediate, simple solutions are not forthcoming. Dr. Lagoo isn’t throwing in the towel. "I think I understand what the problem is," he says of his efforts to set up the assay. "I just need some patient work to get to a point where we can get it to run in our lab consistently."
Likewise, Dr. Shankey, who prior to his arrival at Beckman Coulter spent 13 years in the Pathology Department at Loyola University Medical Center, Maywood, Ill., says the discussions at the meeting, while inconclusive, may have sparked something just as valuable: Instead of lifting attendees above the trees to see the forest, they pointed out the individual timbers standing in their way. That’s exactly what ZAP-70 needs right now.
As he listened to the discussions on gating strategies, antibodies, lysing systems, and so on, Dr. Shankey says, "I just sort of sat and thought, ’Thank God.’ Because flow cytometry’s gone through this about 15 times before in its history. And this time it’s only taken us a year to figure out we need to get to a consensus. So the good news is, the meeting pointed out, at least to the people there, that there are a lot of technical issues. And that means you’re going to need a working group."
Such a group will be headed by Dr. Marti, who volunteered for the job; others at the meeting were quick to offer their services as well. Dr. Marti will review the situation, then propose, with the help of others, a consensus method that people can use to compare against their own protocol.
In the meantime, the meeting scarified what has, until now,
been untrodden ground. ZAP-70 analyses have largely been in the hands of researchers,
for better and for worse. As is often the case, the journey to clinical labs
has been punishing. Labs that have tried setting up ZAP-70 know this firsthand.
"But I don’t think that the published, scientific studies to date have mentioned
the issues we’re struggling with," says Lawrence Hertzberg, MD.
The gap quickly became clear as attention moved from one poster to another.
Dr. Hertzberg, director of hematopathology and oncology services, Genzyme Genetics, Los Angeles, presented his company’s validation data for ZAP-70 evaluation by flow cytometry. The two-site study used a novel four-color direct staining on a variety of specimen types and patient diagnoses in a clinical laboratory setting.
Dr. Hertzberg ticks off one problem after another. Like other ZAP-70 experts interviewed for this article, he introduces his points with the phrase, "that’s the other challenge." By the end of each conversation, there aren’t enough fingers and toes available to count up the entire list of these supposedly singular worries.
For starters, ZAP-70 is an intracellular antigen. Of course, flow labs routinely and reliably assess numerous antigens on intact cells with permeabilized membranes. However, permeabilization for detecting intracellular antigens adds another source of variability and can result in low signal-to-noise ratios.
The literature on ZAP-70 flow cytometry has all shown a shift in fluorescence
intensity when studying cells that are ZAP-70 negative and comparing them to
CLL cells that express the antigen. T cells generally show brighter staining
than CLL cells. Though staining intensity is dependent on the clone-fluorochrome
combination (or combinations) used, in most published histograms, the degree
of shift of CLL cells is minimal, even when an entire population of CLL cells
expresses the antigen.
Most of the research papers then use a percentage above a cursor setting to determine positivity and negativity—yet another problem. Generally speaking, a negative control antibody, in a separate tube, is used to determine the cursor setting. For that to work reliably requires an identical degree of staining in negative cells between the probe antibody—in this case, the antibody that binds specifically to ZAP-70—and the control.
"We found that worked very well when we first did our validation study," says Dr. Hertzberg. Then their troubles began.
The first problem was a lesser degree of staining in the negative control antibody
in subsequent lots. So if the researchers simply used the procedure and didn’t
rely on negative controls, they ran the risk of generating false-positive results.
The problem appeared within a few months of the initial validation study. "Same
source of reagents, same manufacturer," says Dr. Hertzberg. Though he and his
colleagues are accustomed to some week-to-week or month-to-month variations
in the reagents they purchase, it becomes a bigger issue for ZAP-70 because
of the tiny change in staining between positive and negative cells. They’ve
since decided to use internal negative controls, whenever possible, for future
ZAP-70 flow cytometry evaluations.
With ZAP-70, the most important internal negative control is the benign B cells that are present in some, but not all, samples, says Dr. Hertzberg. But the traditional quantity of data collected may be insufficient to detect a good internal negative control in a portion of cases. "That’s the other problem," he says. "The solution is to collect more data. However, if one has a thousand-fold difference between the number of internal negative control cells and CLL cells, it becomes an issue of how to analyze the data properly to get reliable results."
Dr. Hertzberg would like to see tighter quality control of reagents, especially in terms of staining intensity. Initially he and his colleagues tried to obtain the same reagent used in the most recent ZAP-70 paper, published in the New England Journal of Medicine, but even after calling the manufacturer they were left empty-handed. He notes that there seems to be a move to use an Alexa Fluor conjugated monoclonal antibody to detect ZAP-70; its manufacturer, along with several others, also offers a phycoerythrin conjugated antibody. "When we were originally looking to evaluate this particular procedure, we compared both antibodies." The latter—in this case provided by two different manufacturers—provided "somewhat more intense staining," Dr. Hertzberg reports. After a month or two of routine use and storage of the PE conjugate, however, staining intensity began to diminish. By three or four months, it appeared as if the conjugation of the phycoerythrin to the antibody had actually come apart completely, he says, and the PE conjugated antibody no longer worked at all.
"We had some drop-off in staining intensity with the Alexa antibody conjugates that we bought," he says, though the problem wasn’t as great as with the PE conjugates. PE has a significantly higher quantum yield and thus is brighter than Alexa, he adds, which makes a stable PE conjugated antibody highly desirable. "I think we would get more reliable results, and I think the assay would become easier for everyone to do."
A second poster, presented by Antony Bakke, PhD, used the
same four-color assay as the Genzyme group to identify lymphocyte subsets, but
looked at four different commercially available antibodies for ZAP-70.
When Dr. Bakke, professor of pathology, Oregon Health and Science University, Portland, and his colleagues first tried setting up ZAP-70 analysis by flow cytometry, the results were unsettling. Different antibodies gave completely different results. Results also differed markedly depending on the exact way the specimen was analyzed. For a typical clinical lab, the degree of difficulty in performing flow analysis was substantial. "A trained technologist could easily indicate the same patient was positive or negative, depending on only very slight differences," Dr. Bakke says.
They then tried to set up a more robust method that would be applicable across multiple labs and multiple commercial reagents. Their approach involves using a ratio rather than percentage: the amount of ZAP-70 in CLL cells over the amount of ZAP-70 in T cells.
With this approach, each laboratory would need to establish a range of ratios for its particular procedure. Each antibody appears to give a different ratio—in the Oregon study, ratios for one antibody ranged from 0.1 to 0.9; a different antibody, from a different company, exhibited ratios from 0.6 to 2.0. "But any laboratory should be able to decide with their particular antibody and procedure what their range is, and then decide what the upper 40 percent or so of that range is, which would be considered positive for ZAP-70," Dr. Bakke says.
Adding to the difficulties, none of the antibodies have been completely mapped to their binding sites on ZAP-70. Dr. Bakke compares the matter to measuring stem cells with CD34 assays, where the different classes of antibody all bind quite differently and therefore provide different results. Over the past 10 years the flow community has arrived at a consensus about the optimum approach for stem cell measurement. Those who do ZAP-70 will need to do the same, "hopefully in the next year or two," Dr. Bakke says.
The complicated nature of doing ZAP-70 measurements has taken many by surprise, he continues. "Many labs, just like my lab, started to set it up thinking, ’The heme-onc docs want it, it looks like a pretty straightforward thing, the data look good.’ So they started setting it up for themselves. And almost everyone has come to the same conclusion: that they don’t know how to report out the data."
That means the flow community will also need to come to another consensus on whether to use a percentage method for results reporting or some other method, such as the ratio proposed by Dr. Bakke and his colleagues. "I’ve talked to at least five or six different clinical labs where they started setting up ZAP-70, and then decided not to until they learned more," Dr. Bakke says.
There is indeed more to learn, much, much more.
ZAP-70, as it turns out, is about as stable as Liza Minnelli’s marriage. The
protein’s lability is a matter of great interest to Bruce Davis, MD, president
of Trillium Diagnostics, Scarborough, Me. ZAP-70 caught Dr. Davis’s attention
when he directed a flow cytometry laboratory at Genzyme Genetics. In validating
the assay he helped set up there, "We found ZAP-70 is not stable after 24 hours
in a tube of blood," he says. "And it may drop off at different rates depending
on whether you’re looking at T cells or B cells."
This has obvious implications for reference lab testing, and quite likely local labs as well. "One is in a rush, basically, to determine the ZAP-70 expression as soon as the sample hits the door," says Dr. Hertzberg.
At the Long Beach meeting, Dr. Davis presented a poster on interlaboratory correlation of ZAP-70 studies, based on further work at Genzyme done in conjunction with three other labs: Esoterix; Dartmouth-Hitchcock Medical Center, Lebanon, NH; and Northwestern University, Chicago. Each group in turn found normal and CLL samples at their institutions, then split them and sent them to the other three labs; the samples, some 75 in all, were tested the following day, with each group using its own specific analytic method.
"We did get reasonable correlation in terms of the ZAP-70 levels on the CLL samples," Dr. Davis says. "We probably had a correlation coefficient of around 0.75, thereabouts. But it wasn’t as good as we hoped to see." They found even less correlation in the measurements of normal T cells.
Like Dr. Bakke’s group, these four labs were also interested in results reporting. Although each of the key studies has shown a strong prognostic predictability of ZAP-70 measurement, each has used a different technical approach. "Since each of our four labs all had slightly different variations in our methods, we wanted to see what, if anything, we could do to bring out the best way to express results," Dr. Davis says.
They found that fluorescence quantitation, using a bead-based method to calibrate fluorescence intensity, gave better correlation than the so-called percent positive method used most commonly in the literature. (The August 2004 NCCLS document I/LA24-A outlines the various methods for quantitative fluorescence measurements in flow cytometry.)
Of course, reporting percent of positive cell population is the common approach for most flow cytometry studies. But ZAP-70 is a different animal, Dr. Davis and others contend, because typically the entire population of what is being measured, whether it’s CLL cells or T cells, move or act as a single population with expression; what differs is the amount of ZAP-70 expression, whether high or low. "In a really strict sense, reporting the results as a percent positive is an illogical way to look at it," Dr. Davis says. In other words, percent positive is fine for black-and-white answers, but ZAP-70 answers are shades of gray. That’s why he and his abstract co-authors suggest considering a ratio-based approach for reporting ZAP-70 results: fluorescence intensity of the CLL cells to either a negative population or autofluorescent sample.
"Rather than looking at positive versus negative in two-dimensional space, we’re looking at a single-parameter fluorescence distribution of cells collected through Boolean gates. The position of the distribution along the axis is proportional to the ZAP-70 expression," says Marc Langweiler, DVM, PhD, a technical specialist in the Dartmouth-Hitchcock laboratory. While the results were not as cheering as he and his colleagues had hoped, it kept the results reporting issue in the spotlight, where it needs to be, Dr. Langweiler says.
The key ZAP-70 papers in the literature that have used flow cytometry have been published by researchers whose expertise lies in areas other than flow, Dr. Langweiler observes. "It’s sort of been done by people who know cytometry, and the people who believe in fluorescence quantitation are more what I’d like to call—well, the reason I do it that way is because I’m a geek," he says with a laugh. In fact, he predicts that this may be a bump in the road for flow cytometry in general. Percentages and cursors have been part and parcel of flow since day one, he says, but in the last several years fluorescence quantitation has made headway as a possible alternative for ZAP-70 and other mind-bogglers. "I think this is trying to come to the fore of the cytometry community," he says, "and it just hasn’t made it there yet."
But as another observer points out, while ratios may be analytically more satisfying to do, "no one knows what that means clinically. All the clinical data out there is based on whether something is positive or negative and a particular percentage cutoff," says Eric Hsi, MD, head of the section of hematopathology, Cleveland Clinic Foundation.
Unless the flow community reaches a consensus on reporting results, Dr. Davis fears ZAP-70 will go the way of DNA ploidy. "There was, and still is, very strong evidence that DNA ploidy has prognostic value for breast cancer," he says. "The problem is, we ended up with a lot of laboratories with variable experience and variable technical competence doing the assay. It got so confusing that essentially the clinicians lost faith that this measurement really means anything, because they’d get different values from different laboratories and couldn’t reproduce results from lab to lab."
Like Dr. Shankey, Dr. Davis expresses the hope that in the matter of ZAP-70, the lab community will reach a consensus sooner rather than later. The pressure to do so will only continue to build. Local hospitals and regional reference labs are faced with demands to set up the assay, and until now they’ve been trying to do it by perusing the literature. "Unfortunately, there’s going to be a period of adjustment and change until something’s settled on," Dr. Davis says.
Interestingly, the four-site study failed to show a correlation between ZAP-70 expression and CD38 expression. Yet it’s too early to rule out CD38 as a bearer of additional prognostic value, Dr. Davis says.
Dr. Marti says he was surprised by the finding, but urges caution. "On the basis of this poster, you might throw CD38 out, which would be the wrong interpretation." CD38 initially appeared to be a possible surrogate marker for IgVH but has shown to have a discordance of approximately 25 percent. "Everyone was bent out of shape that CD38 wasn’t the perfect marker because of this," Dr. Marti remembers. But similar discordance rates for ZAP-70 are being overlooked today. "It’s the same story all over again," he says.
The fourth poster described a single-institution study which assessed three levels—low, medium, and high— of ZAP-70 expression. "They too were bothered about how difficult it was to interpret positivity versus negativity," says Dr. Marti. "And then, of course, it’s highly variable depending on which internal control you’re using."
This group used directly conjugated antibody clone 1E7.2 conjugated to phycoerythrin. They found a clear decline in staining intensity in 24 hours, both in T cells and background B cells; they also noted the problem of a substantial gray zone, due to heterogenous staining distribution, when placing the cursor on T cells. B cells appeared to have some ZAP-70 expression when the cursor was set relative to the isotype control or T cell control—but B cells biologically lack ZAP-70. "This generated all kinds of discussion about whether B cells really lack ZAP-70, and whether the antibodies cross-react with Syk protein (which is in the same family as ZAP-70)," says one observer.
Four posters and plenty of conversation later, the ad hoc
meeting ended. "They actually had to cut it off," says Dr. Shankey, because
it was on the verge of overlapping with a scheduled event. Otherwise, he says,
"I think it would have kept on going."
Now it will be up to Dr. Marti and amenable colleagues to help keep things going. Most agree Dr. Marti is the right man for the job.
Reached at his offices in Bethesda, Md., Dr. Marti strikes an interesting balance between experience and world-weariness. He’s impressed enough by ZAP-70 to have volunteered to head the consensus effort. ZAP-70 will be a useful test, he predicts. But in his next breath he warns, "We should be careful not to put all our eggs in one basket. FISH cytogenetics is still pretty powerful. I don’t think anyone has really shown that immunoglobulin gene sequencing, or CD38, or ZAP-70 for that matter, is superior to cytogenetics." He pauses before adding, "But what about a flow-based FISH method? That could make a difference, and would certainly be an innovative use of trying to combine assays." Then he circles back with this concluding thought: "I think it’s worth trying to solve this, to make ZAP-70 routinely available."
His starting point will be to review what he considers the five key papers that have lifted ZAP-70 to its current, visible height.
Paper No. 1: an article in the Journal of Experimental Medicine (Rosenwald A, et al. Relationship of gene expression phenotype to immunoglobulin mutation genotype in B cell chronic lymphocytic leukemia. 2001;194:1639-1647), the landmark paper that started it all. Using so-called genomic scale gene expression profiling, the researchers discovered a common molecular signature for CLL that differentiated it from other normal B cells and from other lymphoproliferative diseases. When they took the next step and divided patients on the basis of their immunoglobulin gene mutational status, they were able to discover, and describe, CLL subtype distinct genes, one of which was ZAP-70. This microarray finding was then confirmed by RT-PCR.
The second paper (on which Dr. Marti is an author) was published in Blood last year (Wiestner A, et al. ZAP-70 expression identifies a chronic lymphocytic leukemia subtype with unmutated immunoglobulin genes, inferior clinical outcome, and distinct gene expression profile. 2003;101:4944-4951). This study, which profiled 107 patients, confirmed and expanded the findings of the first paper. In addition to looking at microarray and immunoglobulin gene analysis, the researchers looked at Western blotting and immunohistochemical analysis. Instead of using RT-PCR, they used quantitative PCR, which, according to Dr. Marti, showed excellent correlation between RT-PCR and the microarray results.
Both papers were primarily based on the molecular relationship between microarrays and ZAP-70. The next three papers switch gears and deal primarily with flow cytometric determinations—a good thing, since few labs stock either microarrays or immunoglobulin gene sequencing methods in their pantries, so to speak.
The first of the three, by Crespo M, et al, looked at 56 patients (ZAP-70 expression as a surrogate for immunoglobulin-variable-region mutations in chronic lymphocytic leukemia. N Engl J Med. 2003;348:1764-1765). The researchers showed a correlation between ZAP-70 and both survival and time to treatment for Binet Stage A CLL patients, and their paper was accompanied by a glowing editorial. "When I saw this paper, when I saw the flow cytometric assay they used, my immediate impression was, This is not ready for prime time," Dr. Marti recalls, explaining the ins and outs of the group’s sequential gating method and small scale, among other things. A footnote hints at the method’s difficulty, Dr. Marti says, "but I don’t know how closely people read that. And I think that perhaps the reviewers of the NewEngland Journal of Medicine were not so cognizant of just what a difficult assay that was."
Not long afterward, another paper appeared in the Lancet (Orchard
JA, et al. ZAP-70 expression and prognosis in chronic lymphocytic leukaemia.
2004;363:105-111). Orchard and her colleagues used microarray expression and
a flow cytometric method to look at ZAP-70 expression in 167 patients. Dr. Marti
draws attention to several aspects of the group’s method, starting with its
decision to use 5 x106 cells fixed in para
formaldehyde. "That’s an interesting number," he says—1 x106
would be a more typical approach—"and that means they were probably sensitive
to cell loss that would occur." The group also used a single-color assay—highly
unconventional but nonetheless effective for this study. "But they did validate
it using a T cell marker, CD3 or CD2," says Dr. Marti, "and they eventually
used the remnant T cells as an internal positive control." Orchard and colleagues
used a cutoff of 10 percent, while Crespo’s group used 20 percent. Like Crespo,
Orchard also showed a link between survival and ZAP-70 status, as well as correlation
between ZAP-70 and immunoglobulin gene status. "Their data suggest that ZAP-70
and immunoglobulin gene sequencing, as they correlate with survival, are of
about equal power from a statistical standpoint," Dr. Marti says.
The most recent paper pushes ZAP-70 data to new highs, both practical and esoteric (Rassenti LZ, et al. ZAP-70 compared with immunoglobulin heavy-chain gene mutation status as a predictor of disease progression in chronic lymphocytic leukemia. N Engl J Med. 2004;351:893-901). This is the first time researchers have shown ZAP-70 positivity to be superior to immunoglobulin gene mutational status. Time to treatment for ZAP-70 positivity 20 percent or greater was 2.9 years, versus 9.2 years for patients who had less than 20 percent positivity. "That’s a pretty spectacular finding," says Dr. Marti. It gives clinicians the opportunity to observe more carefully and plan for earlier treatment in such patients, and it will help determine the frequency of office visits.
The Rassenti group, which looked at 307 CLL patients, almost exclusively used peripheral blood mononuclear cells that were cryo preserved. They carried out their analysis on CD19+CD3- cells, using CD19-CD3+ T cells as their internal control, Dr. Marti reports. They did their surface staining first, then fixed, using four percent concentration of paraformaldehyde. "It’s kind of high," Dr. Marti observes, but it may have been necessitated by their use of saponin for permeabilization.
This group used a different clone than either the Orchard or Crespo groups, 1E7.2, conjugated to Alexa Fluor-488 dye instead of FITC. "The Alexa Fluors are a new group of fluorochromes that probably are more stable and show less variation with pH. So that was probably a good technical choice," says Dr. Marti.
In another variation, the Rassenti group essentially flipped the analysis procedure used by the Crespo team. The latter group put positive T cells in the upper right-hand quadrant and tumor cells in the lower right quadrant. Rassenti and her colleagues, on the other hand, set their cursors so there is less than 0.1 percent positivity in the upper right quadrant. "Then they look at CD19+CD3- cells, and whatever number of events that fall into that, those are considered to be positive," Dr. Marti explains. They also use a recursive partitioning method to show that a 20 percent cutoff is useful for determining ZAP-70’s relationship to survival and the need for early treatment.
Reaching for the ibuprofen yet? The variation among each group’s
technical details "can eat you alive," Dr. Marti says. But they can’t be overlooked
by anyone searching for a ZAP-70 consensus.
Like a swimmer completing his laps, Dr. Marti slices through the known world of ZAP-70 and then emerges, dripping with facts and shaking them off to ask some questions of his own. Among them:
- Is there a way to amplify the signal? "It would be fantastic if there was,"
Dr. Marti says.
- Is there a better antibody out there?
- Do laboratories have to purify or enrich? One would hope not. The literature
suggests the answer is yes, depending on the assay. But the issue is far from
clear, Dr. Marti says. Fresh cells, whole blood—either may be equivalent,
according to two of the seminal flow papers, and the third makes no clear
distinction. Then again, as another observer remarks, none of these distinctions
may matter for the average lab. Enriching for B cells, as the Rassenti group
did, may be a smart move, says Dr. Langweiler, "except from a laboratorian
point of view, it’s incredibly impractical."
- Is it time to look at other markers? "Other people have already suggested
that, independent of [the current literature and findings], there may be other,
’easier’ markers that differentiate unmutated and mutated patients," Dr. Marti
- Is ZAP-70 best left in the hands of researchers? "Maybe that’s a legitimate
conclusion," Dr. Marti says. "And that’s fine. It just means they’ll have
to step up to the plate and make this available."
- Are the research labs aware of the problems their methods have caused for
clinical labs? "They should be aware that some laboratories in the U.S. are
not able to reproduce their data. I don’t think it means there’s anything
wrong with the data," Dr. Marti says. "It just means it’s a difficult assay."
Or, as Dr. Langweiler says, "If anyone who’s doing this tells you it’s not
challenging, then they’re not telling you the truth."
Dr. Marti even suggests it may be time for the National Cancer Institute to become involved in ZAP-70 discussions. Thomas Davis, MD, senior investigator in the NCI’s Clinical Investigations Branch, Cancer Therapy Evaluation Program, agrees with the need for a ZAP-70 workshop, but says it should encompass other prognostic markers in CLL.
In the meantime, what’s a lab to do?
Waiting around for other prognostic markers is probably not the answer. In fact, one more marker may make the field look like the 2nd act of "Aida," with the audience questioning the director’s crowding the stage with yet more horses or, heaven help us, elephants. In the intellectual history of CLL, prognostic markers may have reached their apogee. "We now have to figure out a little bit more of the underlying mechanism," Dr. Marti says.
If it is indeed time for labs to dance with the ones that brung them, what does that mean, given all the questions and uncertainties whizzing thru the air?
First, ZAP-70 should not be done in isolation. Dr. Marti suggests turning to one of the methods that first brought ZAP-70 to prominence: quantitative RT-PCR. "That’s one method that labs should have in their back pocket, or consider having," he says, particularly for purists who want to adhere to the developmental logic of ZAP-70 assays. The other option would be using Western blotting or IHC. Regardless of the method chosen, however, there should be another method on the shelf, done simultaneously with whatever flow method the lab may be trying to set up. "If you have any doubt about the result you’re reporting out in flow cytometry, then you have to do one of two things," Dr Marti explains. "Don’t report it—and that means you really can’t offer the assay for sale. Or, have a method that you know can be used to support it."
Each of the three methods—with the possible exception of RT-PCR—has a drawback, he concedes. "Some of this discordance is real. It’s not technical. It’s biological. And there’s nothing we can do about it. But to the extent that you can protect yourself, as a pathologist, I would recommend having one of these three methods for the time being, while you worked on flow."
His other recommendation pertains directly to flow. Despite his admiration of the one-color method used in the Orchard paper, he says, "I could not, in all honesty, recommend a single color." Three colors, in his opinion, is the minimum, and four colors would be even better. Yes, he knows that means added complexity in terms of controlling for multiple colors, and he’s aware that neither Crespo nor Rassenti described their compensation controls. (That’s another can of worms, and he’s happy to discuss the pros and cons of intuitive compensation versus software-based approaches. Suffice it to say that his preferred method is to collect uncompensated data, with compensation controls, followed by software-based analysis.)
Some other suggestions, for what they’re worth:
- Run a normal donor control to establish fluorescent range. The scale of
fluorescence on the positive end is anchored by normal T cells and NK cells,
the negative end by normal B cells. "That’s your scale, whether you like it
or not," says Dr. Marti. "I don’t know of any way to amplify that scale."
- The Rassenti group ran a CLL sample that was strongly positive and a CLL
sample that was weakly negative in addition to its normal controls—a
more rigorous approach that clinical labs may want to emulate, Dr. Marti says.
The other research groups tended to use the normal remnant T cells that were
still present as a positive control but were unclear as to what their negative
control was—they may have mixed, in their minds, negative B cells in
normal donors with negative ZAP-70 CLL cells, Dr. Marti hazards. "It’s OK,
but it’s not so rigorous," he says.
- Consider using a positive cell control line, specifically the Jurkat cell
line. Not everyone agrees on this, says Dr. Marti, recalling the vigorous
discussion at the ad hoc meeting in Long Beach. Frankly, the majority of clinical
diagnostic flow labs lack the equipment needed to grow a cell line. But it’s
possible, he says, that this could become part of a send-out service. David
Barnett, PhD, MRCPath, of the University of Sheffield Academic Department
of Haematology, Royal Hallamshire Hospital, in Sheffield, England, is trying
to prepare a stabilized whole blood ZAP-70 positive control, which Dr. Marti
says would be a "welcome addition." And some labs—notably Dartmouth
and Esoterix—are using microbead standards, which he says should make
it easier to develop a calibration curve.
All of this presupposes flow is the way to go. While many
observers suggest the problems with flow are only compounded in IHC, not everyone
One dissenting voice belongs to Madeleine D. Kraus, MD, assistant professor of pathology/laboratory medicine and director of the immunohistochemistry lab, University of Chicago, where ZAP-70 is being eased into clinical use.
The lab uses a two-color immunostain that combines ZAP-70 expression with that of Pax-5, a B cell marker. Pax-5 is a transcription factor with a nuclear pattern of expression, which pairs well with ZAP-70 and its membrane pattern of expression, says Dr. Kraus. This allows the lab to determine whether the B cells are in the bone marrow, in the peripheral blood, or in the involved lymph node. "That’s important, because ZAP-70 can also be expressed and is expressed in T cells, benign and neoplastic," she notes. They report a case as positive only when both markers are present—the brown stain for Pax-5 and the red stain for ZAP-70. (Dr. Kraus discusses the double-stain procedure, in a different but analogous diagnostic setting, in an article she co-authored: Kraus MD, et al. Lymphocyte predominance Hodgkin’s disease: the use of bcl-6 and CD57 in diagnosis and differential diagnosis. Am J Surg Pathol. 2000;24:1068-1078.)
Dr. Kraus concedes that double immunostains are technically difficult for many labs, but their ability to target specific issues in a highly refined manner make them worthy of consideration. "In this particular situation, for instance, we don’t really care what the T cells are doing in a bone marrow in a patient with a history of CLL; we care what the B cells are doing," she says. Since the T cells will be positive for ZAP-70 and resemble small lymphocytes, it’s difficult to look at bone marrow that’s only partially involved and to discern whether the positivity relates to the T cells or the B cells. At the end of the day, even the best single-marker studies will not answer that question, she says. "ZAP-70 is a very important marker, and possibly five years from now it will replace CD38," she says. "But it needs to be applied in a logical, biologically meaningful way. And for me, that means knowing not just that there are ZAP-70 positive cells in the patient’s bone marrow, but that they’re part of the patient’s leukemic clone."
Flow methods have their own problems in addressing the ZAP-70 status of CLL cells, she says. "It’s easy to look at that pattern of expression because it’s a membrane as well as a cytoplasmic marker, but it’s actually very difficult to exclude the T lymphocytes from study without some fairly detailed double markers. That’s why we’re turning toward immunohistochemistry as a way of independently addressing that question." IHC also enables the lab to clear the permeabilization hurdle that many flow labs are tripping over. And while she’s well versed in the intricacies of double immunostains—she’s been doing them since 1977—she notes that more manufacturers are starting to sell double antibody detection kits, which could eventually make the method a reasonable choice for more labs.
Dr. Kraus says her clinical colleagues have welcomed the ZAP-70 analyses offered by her lab. "It’s been on the academic radar screen since 1999, and they’ve been hoping we can do something that’s faster and less expensive than gene sequencing." U of C patients also tend to be highly educated about their diseases, she adds, which has further boosted the demand for ZAP-70 testing. But she’s careful to emphasize that ZAP-70 can’t stand alone; at U of C, it’s joined by CD38 flow analysis and a cytogenetics workup for 17p and 11q abnormalities as well as trisomy 12 and 13q abnormalities.
Dr. Kraus strikes a new note in the ZAP-70 discussions—calm.
Without glossing over the technical challenges of ZAP-70 evaluations, she manages
to convey a sense of being in control. Indeed, there are those who are managing
to get ZAP-70 up and running.
Such is the case at the Cleveland Clinic, where the majority of cases are done by flow, though ZAP-70 is occasionally done by IHC. "It really depends on what kind of tissue we have available," says Dr. Hsi—IHC is the choice when flow material is not available. And a double immunostain is important when there are evenly mixed populations of B cells and T cells, he says, such as a bone marrow tissue section.
What’s his secret? He has none, unless you consider "keep trying" to be a password of sorts.
"I think everybody has the challenge that either the antigen is not highly expressed, or else the reagents aren’t as robust as you would like for detecting ZAP-70 by flow cytometry," says Dr. Hsi. His lab, like others that are reporting reasonable success with ZAP-70 flow, tested several manufacturers’ reagents before finding one they felt was both relatively easy to use from a technical standpoint and showed good separation between normal B and T cells and normal NK cells. "I think if you look at a variety of reagents with a variety of protocols and settle on the one that works best in your laboratory, there are enough internal controls to give you at least some confidence that you’re able to have an informative assay. And that’s about the best you can do with the current state of affairs," he adds.
The real problem, he says, is the lack of a gold standard by which labs can validate their method. "Most clinical laboratories don’t do clinical validations of this test—it takes five, 10 years, so it’s really not practical," he says. That leaves only one tenable option, at least for now, he says—see how the lab’s own numbers match up to the positivity rates described in the literature. Dr. Marti suggests that an external proficiency testing program also be considered, or "comparing the local method to a consensus reference protocol as envisioned by CCS members."
ZAP-70 also becomes more feasible when labs can accept—and get their clinicians to accept— that they won’t always be able to provide an answer.
At Dartmouth-Hitchcock, says Dr. Langweiler, ZAP-70 is included in the antibody panel used for CLL, but it’s reported separate from the other immunophenotypic markers (and is not charged for) because the ZAP-70 assay is not yet an analyte-specific reagent technique; a disclaimer to this effect is included in the report. "We’ve met with our oncologists and discussed the pitfalls of the current state of the art, and how and why we present our results the way we do." The clinicians are aware, he says, that the results need to be considered in the context of other laboratory determinations and clinical presentation.
Clinicians will reconcile themselves to nonresults if labs can convey the technical difficulties of the test. "The fact is, people want you to do this test," Dr. Hsi says, even though a percentage of cases will be indeterminate, at least for the time being.
J. Brice Weinberg, MD, professor of medicine, Duke University School of Medicine, reports experiences similar to Dr. Hsi’s. Finding the right antibody and permeabilization technique has been a matter of considerable trial and error. "Some antibodies are just better than others," he says, "and I think that’s what held a lot of people back initially, not having the ’best’ antibody. That can create a lot of problems—it certainly did for us." Even after settling on an antibody, however, their work wasn’t done. "You have to fiddle with it. It’s very susceptible to background and the gating, and you have to have a real expert playing with the machines," he says.
"Our best luck has been with doing immunoblots," he says. "We get more consistent findings, and then we can quantify the density of the reaction." Immunoblot isn’t an easy answer, though. "It’s more complicated than flow cytometry and it takes more time," Dr. Weinberg says. On the plus side, "It’s accurate when it’s done right." Each sample is done by flow as well as immunoblot, though he’s unable to comment on correlation because the data have not been analyzed. The lab also does sequencing to assess immunoglobulin heavy-chain gene mutation status. So far they’ve restricted their data to research, not clinical, use.
Dr. Lagoo at Duke University knows there are a handful of labs that are succeeding with ZAP-70. He’s talked to some of them—including his Duke colleague, Dr. Weinberg—looking for ideas. "But unfortunately each of those labs has a different protocol. And unless it works in your hands, it’s of no use," Dr. Lagoo says.
In fact, Dr. Lagoo thought he was one of the lucky few early on, when he first tried setting up the assay using the standard protocol his lab used for other cytoplasmic markers. "We saw a lot of cells positive, which made us happy, because that means at least that your stain is working," he recalls. Then they used the protocol on their samples of CLL cells, at which point the good news became too good. "Everything was positive, 100 percent. Getting positive results is apparently much easier than getting cells that are negative to stain negative," he says, laughing.
So he keeps at it. One investigator told him he used the antibody at a much lower dilution than recommended by the manufacturer, and that he blocked nonspecific binding using normal mouse serum. Dr. Lagoo has also ordered a rabbit monoclonal antibody, which, it has been suggested, may have better affinity than mouse monoclonals. Other items on his to-do list include testing several ways of permeabilizing the cells, looking at various methods of blocking nonspecific binding before and after permeabilization, changing incubation temperature and length, and placing CD19 and CD5 markers on the surface to pick up abnormal cells. He’s also going to explore a method that sidesteps the problems associated with gating, instead using the median channel for positive control T cells and for ZAP-70 positive CLL cells. "You divide your CLL cells by the T cell median channel, and if the CLL cells are positive, they should have approximately the same amount of ZAP-70 fluorescence as the T cells," he explains.
It’s a long list. "It means a lot of work, and who knows if we’ll be successful," Dr. Lagoo says. "But that’s where we are."
It’s probably where most labs should be. If anything came across loud and clear
at the CCS meeting, he says, it was that ZAP-70 cannot be ignored—the
data supporting its use as a prognostic marker are just too compelling, and
clinicians and patients, understandably, are knocking at the door.
Karen Titus is CAP TODAY contributing editor and co-managing editor.