How clinical and histopathologic workups come together for a definitive diagnosis
Sherrie Rice
June 2024—Idiopathic multicentric Castleman disease, which is driven by a cytokine storm with an unknown cause, is a difficult diagnosis and one that’s often delayed, owing to the disease’s rarity and nonspecific symptoms.
“Patients often bounce around for months, or even years, to different specialties, based on how they present, before they are diagnosed,” said Jadee Neff, MD, PhD, assistant professor of pathology, Duke University Medical Center, in a CAP TODAY webinar in February made possible by a special educational grant from Recordati Rare Diseases.
In her presentation (available at captodayonline.com), Dr. Neff explained that an idiopathic multicentric Castleman disease (iMCD) diagnosis must include histopathologic lymph node features consistent with the iMCD spectrum and is best assessed with an excisional lymph node biopsy.
Castleman disease is an umbrella term that refers to a heterogeneous group of disorders that share histopathologic features of the lymph nodes. It can be divided into two groups: unicentric and multicentric. Unicentric refers to diseases that involve one lymph node site; there could be a couple of lymph nodes involved, but it is one site. Multicentric Castleman disease refers to multiple lymph node sites that are involved at the same time.
Multicentric Castleman disease can be further divided into groups based on the etiology. One is human herpesvirus-8 (HHV-8)-positive MCD, which is often but not always associated with HIV infection. Another is POEMS-associated MCD. Patients with POEMS, a plasma cell-related condition, present with polyneuropathy, organomegaly, endocrinopathy, monoclonal gammopathy, and skin changes. All other MCD is lumped into idiopathic, or iMCD. “Currently, iMCD accounts for approximately 33 to 58 percent of all published multicentric Castleman disease cases,” Dr. Neff said.
The annual incidence of iMCD is 3.1 to 3.4 cases per million in the U.S., and the overall prevalence is 6.9 to 9.7 cases per million, according to a retrospective U.S. health claims analysis of 2006–2020 data (Mukherjee S, et al. Blood Adv. 2022;6[2]:359–367).
“This may seem fairly low, but the one caveat of this study is that the diagnostic criteria for Castleman disease were not firmly established until 2017. So it is possible there are cases that were missed in this retrospective study. But it is important to diagnose these cases because these patients do have poor outcomes,” Dr. Neff said.
Patients with iMCD are hard to identify, she said. The disease can occur in people of any age and can affect any patient population. The symptoms can mimic infections, autoimmune diseases, and malignancies. The patients can present variably. Some are asymptomatic with a few laboratory abnormalities; others have some combination of fatigue, edema, weight loss, night sweats, ascites, fever, or enlarged liver or spleen. Some patients present with multiorgan failure.
The underlying cause of the cytokine storm that drives iMCD is unknown, but three major hypotheses exist: an uncontrolled infection, a germline mutation leading to an autoimmune or inflammatory response, or somatic mutations in monoclonal lymph node cells that lead to uncontrolled cytokine secretion. Studies are underway to identify the etiology, Dr. Neff said. “Regardless of the etiology, iMCD is caused by a cytokine storm,” driven by an uninhibited dysregulation of IL-6.
Patients with iMCD have increased VEGF production, which leads to the formation of blood vessels and increased blood supply to tumor, “and that’s what we see on these lymph nodes,” she said. Increased plasma cell and B-cell overgrowth leads to the lymphadenopathy and organomegaly. Also seen: increased T-helper-2 (Th2) cells and markers of inflammatory response (increased erythrocyte sedimentation rate, C-reactive protein, immunoglobulin G, anemia, deep vein thrombosis).
Diagnostic criteria developed by an international panel and published in 2017 made the diagnosis of iMCD “a little more straightforward,” Dr. Neff said.
The two major criteria, both of which must be met, are multiple enlarged lymph nodes (≥1 cm in short-axis diameter in two or more lymph node stations) and histopathologic lymph node features consistent with the iMCD spectrum (Fajgenbaum DC, et al. Blood. 2017;129[12]:1646–1657). The 11 minor criteria are in Fig. 1. Two minor criteria must be met for a diagnosis of iMCD, and one of them must be a laboratory criterion. Also in Fig. 1 are the exclusionary criteria. “Diagnosis of iMCD has become more understood in the community,” Dr. Neff said, “such that now it is included in its own chapter or section of the fifth edition of the WHO.” The fifth edition specifies that the histopathology must contain at least grade two or grade three regressed germinal centers or plasmacytosis.
“Pathologists are essential in making this diagnosis,” she said. “This is not something that can be diagnosed just on the clinical side. If we don’t see the pathology or if we can’t make the call that we see features consistent with Castleman disease, they can’t make that diagnosis and the patient can’t receive the treatment.”
The pathology itself can be variable, Dr. Neff said. “We have a range of histologies, from lymph nodes that are characterized by primarily regressed germinal centers all the way to pathologies that are characterized by hyperplastic germinal centers and abundance of plasma cells.”
An excisional biopsy is key to making the diagnosis, she said. “Some groups do needle core biopsies, and my understanding is that they do several needle core biopsies, so it’s never just one biopsy. But, in my personal experience, I prefer to see an excisional biopsy, and I think that’s the best way to look at the entire histology of the lymph node.”
The scoring criteria proposed by Fajgenbaum, et al., in the Castleman Disease Collaborative Network article published in Blood include assessing and grading five histologic features in the lymph node. Definitive grading criteria with objective cutoffs have not yet been established, Dr. Neff said, so grade assignment is left up to each pathologist. The first histologic feature is regressed germinal centers, with grading on a scale of zero to three (Fig. 2A). “They don’t have any hard cutoffs on what makes a zero or a three versus a two versus a one, but I can tell you in my practice that I tend to use quartiles for this parameter,” Dr. Neff said. If 25 percent or fewer of the germinal centers are regressed, she grades it a zero; 25 to 50 percent, a one; 50 to 75 percent, a two; and 75 percent or greater, a three. “I’ve spoken with a few other hematopathologists just through networking, and others use a similar quartile system.”
The second histologic feature is follicular dendritic cell prominence, also graded zero to three (Fig. 2B). Dr. Neff calls something a three if the follicular dendritic cells tend to take up the entirety of the germinal center. If she can see them clearly (“jumping out at me,” she says) at a 4× objective, she will grade it a three. Follicular dendritic cells are a little less prominent in grade two, “not necessarily covering the entire germinal center but still pretty prominent,” she said. If she can’t see them at all or if she has to “hunt around” for them at high power, it’s a grade zero. A grade one would be “not quite zero but not quite two.”
The third histologic feature to score is vascularity (Fig. 2C). “Again you’re looking at the prominence of the vessels within the lymph node, and a grade three is very easy. This is where you’ll see the characteristic piercing of the germinal center by the vessels—that ‘lollipop’ sign.”
“If the vessels are grade two,” she continued, “you might have to look a little harder, go a little higher magnification, but they definitely are more prominent than a grade one where you might see only one or two vessels, or a grade zero where you don’t see the vessels overtly.”
Hyperplastic germinal centers are the fourth histologic feature to grade (Fig. 2D), and here too Dr. Neff uses quartiles. If more than 75 percent of the follicles have hyperplastic germinal centers, it’s grade three; if 50 to 75 percent, grade two; if 25 to 50 percent, grade one; and if less than 25 percent, grade zero.
The fifth histologic feature to score is plasmacytosis (Fig. 2E). “This one can be the most tricky, if you’re not used to seeing a lot of lymph nodes and you’re not used to seeing how many plasma cells are present in a normal lymph node,” Dr. Neff said. Step No. 1 for her is to break it up into grade zero (no plasma cells or rare plasma cells seen) and grade three (sheets of plasma cells seen in the interfollicular areas). “If you’re seeing the interfollicular areas packed with plasma cells, I would give that a grade three. For me, a grade two would be where I might see focal aggregates of plasma cells. Maybe it’s not sheet-like, but I see aggregates here and there, and plasma cells are fairly easy to see in between the follicles.” In a grade one plasma cells are present, more than grade zero, but “I’d have to hunt around to see them.”
iMCD lymph node pathology, then, tends to be one of two types. One is hypervascular, which tends to show follicular dendritic cell prominence, regressed germinal centers, and hypervascularization without plasmacytosis. “Lollipop” signs and “budding” or “twinning” of follicles (when two germinal centers share the same mantle zone) are common in the hypervascular type of iMCD. On the other end of the spectrum is the plasmacytic-type pathology, which tends to be characterized by plasmacytosis and hyperplastic germinal centers. “If you have mixed features, some plasma cells and regressed germinal centers, this would be called mixed pathology,” Dr. Neff said.
Making the diagnosis on the pathological level can be challenging, she cautions. “The pathology isn’t specific, so you can see something that looks perfectly like Castleman and yet if the patient has a diagnosis of rheumatoid arthritis or lupus, you can’t make the diagnosis of Castleman disease because lupus and rheumatoid arthritis could have lymph nodes that look similar. But one of the things we can do, at least as pathologists, is rule out some of the known mimics,” one of which is lymphoma. “If you get your lymph node, you can do your flow cytometry, or if you see plasma cells, you can get your kappa and lambda stain and rule out monotypic plasma cells. Anytime I get a lymph node that looks like it could be Castleman disease, I always get HHV-8 stain to make sure it’s not HHV-8 related.” She also gets EBER stains.
If she sees the plasma cell type with a lot of plasmacytosis, she orders an IgG4 to make sure it’s not IgG4-related disease. “On the histologic level, if you can rule out all those other entities, then you can move forward with your annotation of it being Castleman disease-like or Castleman disease.”
If a pathologist is suspicious and maybe has a small biopsy or only a needle core biopsy, it’s appropriate to request another biopsy or to tell the ordering clinician that an entire lymph node excision is needed, she said. “It’s okay to say that, and it’s important to say that, to be able to make the diagnosis.”
Two levels of certainty are accepted in diagnosing iMCD, and this applies to clinicians and pathologists, Dr. Neff said. “You can be definitive about it if you meet all the major and minor criteria and you’ve excluded the exclusionary criteria.” A diagnosis of probable iMCD can be made if the major and exclusionary criteria have been met but not the minor criteria.
TAFRO is an iMCD subtype that has a severe or aggressive presentation, so it’s on the far end of the spectrum. TAFRO patients present with thrombocytopenia, anasarca or ascites, reticulin fibrosis in the bone marrow, renal dysfunction, and organomegaly. Patients with TAFRO have severe disease and a higher likelihood of death than patients presenting without TAFRO.
Dr. Neff shared one of her cases to illustrate how a pathologist can help make the diagnosis of iMCD
(Chu P, et al. Rheumatologist. Published Oct. 13, 2021. https://bit.ly/Rheumatol-CD). The patient was a 19-year-old woman found at home unresponsive following a seizure. She was hypotensive and febrile at the time. She was admitted to the ICU and intubated and started on vasopressors and antibiotics. Minimal medical history was available except for a recent diagnosis of systemic lupus erythematosus from an outside hospital and information pointing to a healthy childhood.
Six months prior to this presentation, the patient was 36 weeks pregnant and admitted to the outside hospital with weight loss, joint pain, and a rash. At the time she had elevated bile acids. They gave her betamethasone, she had a C-section, her symptoms resolved, and she was discharged. Then two months prior to the admission at Duke, she was again admitted to an outside hospital for altered mental status and abdominal pain.
At the time, she was found to have hemolytic anemia and proteinuria. She had a renal biopsy, which showed podocyte effacement, and at that time they gave her prednisone; her symptoms improved and she left. “So this is where the working diagnosis of systemic lupus was made, based on these presentations at the outside hospital,” Dr. Neff said.
In the ICU at Duke, she had an initial workup and abnormal laboratory values, including pancytopenia. She had elevated creatinine and sedimentation rate, very high ferritin, high triglycerides and LDH, and low complement and haptoglobin. Her blood cultures were negative and her CSF was normal. The CT scan of her head was read as consistent with PRES (posterior reversible encephalopathy syndrome), and the CT of her abdomen and pelvis showed widespread bulky lymphadenopathy. The differential diagnosis was systemic lupus erythematosus with macrophage activation syndrome versus lymphoma.
Because of the suspicion for lymphoma, an excisional lymph node biopsy was performed (Fig. 3). The overall lymph node architecture is preserved. A CD21 stain (Fig. 4) highlights follicles not readily apparent on H&E. “When I do a CD21 stain and find surprise follicles, sometimes that’s a clue there may be regressed germinal centers.” On higher power (Fig. 5) “we do see this characteristic onion skinning of a regressed germinal center and a bit of follicular dendritic cell hyperplasia,” Dr. Neff said. There is also a lollipop sign—“a vessel that’s piercing the germinal center.” In another area (Fig. 6) vessels are fairly prominent and easy to identify. Plasma cells are increased in some of the interfollicular areas (Fig. 7) though they do not form sheets. Flow cytometry did not identify any monotypic B cells or phenotypically abnormal T cells. The kappa and lambda IHC stains were polytypic, IgG4 was not increased, and HHV-8 and EBER stains were negative (data not shown).
Low power (20×) H&E image of lymph node biopsy showing overall preserved lymph node architecture. [Courtesy of Jadee Neff, MD, PhD.]
Medium power (100×) H&E image (left) and CD21 stain (right) of lymph node. Several more follicles are evident using the CD21 stain than were obvious in the H&E stain. [Courtesy of Jadee Neff, MD, PhD.]
Medium power (200×) H&E image of a single follicle showing prominent follicular dendritic cells, mantle zone “onion skinning,” and a single vessel piercing the germinal center (“lollipop” sign). [Courtesy of Jadee Neff, MD, PhD.]
Medium power (200×) H&E image shows several prominent vessels. [Courtesy of Jadee Neff, MD, PhD.]
High power (400x) H&E image of an interfollicular area shows increased plasma cells. [Courtesy of Jadee Neff, MD, PhD.]
Pathologists can report Castleman-like findings with varying degrees of certainty.
If the clinical team is suspicious of Castleman disease and knows that the exclusionary and minor criteria have been met and is waiting only on histopathologic confirmation, and if there are sufficient histopathologic features to qualify, Dr. Neff will call it “Castleman disease” on the top line of her report. If, however, the clinical team is less certain, perhaps Castleman disease is not on the team’s radar or all the minor criteria have not been investigated, but she sees pathology that is a slam-dunk case for a Castleman-like picture, “I will say something that’s a little less definitive but still up in my top line,” she said, “something like ‘benign lymph node with Castleman-like features.’” In her comment she would include a reference to the 2017 article so they could look up the rest of the minor and exclusionary criteria to see if they could make the diagnosis clinically. On the other hand, if she knows the exclusionary or minor criteria were not met, or if the histology is borderline, her language is more descriptive, perhaps “follicular hyperplasia with polytypic plasmacytosis” or “angiovascular lymphoid hyperplasia.” In her comment, she might have a slightly longer list of potential etiologies, and Castleman could be one of them. “But I wouldn’t name it Castleman [in the top line] if I were not sure,” she said.
In the case of the 19-year-old woman, the histology was “fairly supportive, but at least according to what we knew at the time, the patient had a pretty secure diagnosis of lupus.” Dr. Neff was descriptive in her report: lymph node with dermatopathic changes, regressed follicles, plasmacytosis, and focal necrosis. In her comment, she wrote: “The morphologic findings are compatible with systemic lupus lymphadenopathy. Although some features of the lymph node are ‘Castleman-like,’ the patient’s history of SLE precludes a diagnosis of multicentric Castleman disease.”
A phone call came the next morning. The rheumatology team and hematology-oncology service were on the line. They said, “We don’t think this is lupus. We know she came in with an outside diagnosis, but we’re not thinking that anymore, in part because she’s not responding to high-dose steroids and anakinra for lupus and macrophage activation syndrome, which is what we thought she had.” And they asked, “Can you reevaluate for iMCD?”
Dr. Neff had scored the regressed germinal centers as a three, the follicular dendritic cell prominence as a two, vascularity as a three, hyperplastic germinal centers as a zero, and plasmacytosis was mildly to moderately increased—one to two. “I didn’t see sheets, but I definitely saw more than normal background.”
“Because I saw some plasmacytosis and some regressed germinal centers, I said this was mixed pathology. I didn’t call it Castleman because the clinical team was still deciding how much of the minor criteria were met.” She amended her report and wrote a new comment that said, “The predominant morphologic features in this case are compatible with a diagnosis of mixed-type Castleman disease.” She provided the 2017 reference so they could make sure they met all the other criteria.
They did additional workup, and the patient met both of the major criteria, several of the minor criteria, and all of the exclusionary criteria. Her final diagnosis, combining the pathological and clinical workups, was iMCD. The physicians halted the anakinra and started her on siltuximab (Sylvant), which is an anti-IL-6 therapy. Her pancytopenia and inflammatory markers improved, her symptoms improved, and she was discharged. (Hematology laboratory tests are recommended before each dose of Sylvant therapy for the first 12 months and every three dosing cycles thereafter.)
If not treated appropriately, patients with iMCD have a 35 percent mortality rate within five years of diagnosis.
The disease is progressive and characterized by flares, “and each time they come back, they’ll be a little bit worse,” Dr. Neff said. In a population-level U.S. health claims analysis, patients with iMCD were found to have significantly higher prevalence of several malignancies, compared with a non-iMCD matched cohort (Mukherjee S, et al. Leukemia. 2022;36[10]:2539–2543).
Once patients are properly diagnosed and can begin treatment, their re-presentation to hospitals and emergency rooms declines. Serious comorbidities can be lessened in severity or avoided. If left untreated, eventually patients will have multiorgan dysfunction and death.
“If we see it on the pathology side, even if the clinician isn’t thinking of it, it’s important that we at least mention it in our comments,” Dr. Neff said, “maybe provide a reference for them just to get it on their radar, so they can work up the patient appropriately.”
Sherrie Rice is editor of CAP TODAY.