Round 24: A Story of Persistence

By: Rebecca Manno, MD

Johns Hopkins University School of Medicine

Dr. Manno has no significant financial interest or relationships to disclose.

Release Date: April 8, 2010
Expiration Date: January 1, 2011


In the medical world, persistence typically looks like the workup below.  But, in general it means not giving up when the diagnosis is less than obvious.


  • Discuss an interesting case with a common presentation…but an unusual outcome.

A 78-year-old man from New Jersey presented at Hopkins in July of 2007 for fever of unknown origin.  His family gave a history of functional decline for the proceeding 2 years; in the 6 months prior to transfer, his decline had been quite precipitous.  He complained of overall generalized fatigue and weakness.  His family outlined a substantial cognitive decline and he had some mild headaches.  In April 2007, he began to have daily fevers of 100 to 101 degrees.  His wife documented all of his signs and symptoms in a journal.  At this time an outpatient workup was started, and in May 2007, he was admitted to Thomas Jefferson for the workup of the fever. His workup was quite extensive and it included blood cultures and AFB cultures, which were negative.  His HIV was negative, his sedimentation rate and CRP was elevated, and these were persistently elevated throughout his workups.  He had a colonoscopy and an EGD, both normal. He also had biopsies, which were normal.  He had a muscle biopsy because of generalized weakness, it was read at Hopkins, and it was not particularly revealing.  He also had a bone marrow biopsy at Thomas Jefferson; read at Hopkins, also not particularly revealing. Bilateral temporal artery biopsies were obtained, and they were purported as negative. At this point, he was started on Prednisone (40 mg) because there wasn’t a diagnosis, and he was being treated presumptively for giant cell arteritis (GCA).

In June 2007, he was continued on Prednisone for the presumed diagnosis of GCA. He had some functional improvement, he would have good days and bad days.  A good day consisted of him getting out of bed and helping with household chores.  A bad day would consist of him staying in bed most of the day.  His family definitely noted an improvement in his symptoms with the steroids, but as the steroids were tapered, the fevers started to return.  He was admitted to a hospital in Atlantic City and a second bone marrow was performed.  This also was read later at Hopkins and revealed nothing.  In July 2007, the patient was transferred to Hopkins for further workup, and the family was told that he needed a good rheumatologist.

On his arrival to Hopkins, the patient was exhausted, complained of severe fatigue and mild headache.  He was absolutely confused and complained of general weakness.  He had daily fevers at Hopkins, anywhere from 100.2 to 101 degrees.  He wasn’t eating much, and he was still on Prednisone (10 mg) at this time.  He did not complain of any new rashes or skin findings, oral ulcers, chest pain, Raynaud’s, paresthesias or numbness.  His history was notable for an approximate 40-pound weight loss over 6 months, and he was also complaining of a dry cough.

Past medical history

  • atrial fibrillation, but he was status post ablation, and he had a pacemaker in place.
  • He was on anticoagulation
  • He also had hypothyroidism and acid reflux.


  • prednisone 10mg,
  • folate,
  • synthroid,
  • protonix,
  • sotalol

No Allergies
Family History:

  • negative for rheumatic diseases

Social History:

  • married,
  • 3 adult children (one is physician in NJ)
  • no tobacco
  • no alcohol
  • no drugs
  • retired from sales
  • recently moved to NJ from Florida

Physical Exam

On physical exam, the patient’s T-max was 101, heart rate was 100, blood pressure was 100/60, respiratory rate was 12, and his 02 saturation was 99% on 3 liters nasal canula.  He was not on oxygen at home.  He was uncomfortable and ill-appearing.  His temporal arteries were normal on exam.  There was no lymphadenopathy on exam, and his neck veins were normal.  He had a 2 out of 6 systolic flow murmur.  He didn’t have any rubs or gallops.  There were no other signs or symptoms of heart failure.  There were crackles at his bases bilaterally on lung exam.  His musculoskeletal exam showed there was no synovitis.  He had Heberden’s nodes on his distal interphalangeal.  On dermatologic exam, there was trace livido on the plantar aspect of his left foot.  Otherwise, an extensive examination of his skin did not reveal any rashes.  On vascular exam, his pulses were symmetric, and he did not have any bruits.

On neurologic exam, he had waxing and waning levels of consciousness.  He was oriented to himself, but not the month, the President or the year.  His speech was incredibly slow and labored.  He would only respond to questions 50% of the time.  He was minimally verbal and he had a lot of word-finding difficulties.  His strength, once you could engage him, was essentially normal, but it took a lot of effort.  He was able to ambulate, but he was incredibly unsteady.  His gait was slightly wide-based, and his reflexes were normal.

Laboratory data

His CBC was abnormal with a white count of 3.8, hematocrit of 28.3 and platelets of 97,000.  His creatinine was 1.2.  His urinalysis had no protein and no red blood cells.  He had a multitude of other studies; markedly abnormal, which included an albumin of 1.7, an elevated LDH of 2138 and an elevated ferritin of 2303.  He also had a number of serologies sent.  He had an elevated rheumatoid factor at 86 and, once again, elevated inflammatory markers with a sedimentation rate of 67 and a CRP of 7.9.

Imaging studies

The patient had a multitude of imaging studies that included transthoracic echo, which was normal except for some trace mitral regurgitation, aortic valve sclerosis and a tiny pericardial effusion.  A CT of his chest, abdomen and pelvis showed mild bilateral pleural and pericardial thickening, and small pleural and pericardial effusions.  It also showed bilateral ground-glass opacities, mild splenomegaly and left adrenal enlargement.  There was no other lymphadenopathy.  He had a PET/CT, which showed diffuse metabolic activity in both lungs and mild metabolic activity in the left adrenal gland.  He had a CT of the head that showed nonspecific periventricular white matter hypodensities; MRI of the brain with gadolinium showed small areas of subacute ischemia in the periventricular white matter and age-appropriate atrophy.

The patient had a multitude of invasive studies, many of these upon the request of rheumatology, which included a lumbar puncture that had 1 white cell, 23 red cells, an elevated protein of 88 and glucose of 57.  Infectious workup from the CSF was negative.  He had a bronchoscopy because of the ground-glass infiltrates that was negative for infectious workup. Cytology was negative, and transbronchial biopsies were not obtained.  He had a PPD that was negative, and whole blood peripheral flow was done that showed no abnormal cells.

Case fundamentals

  • Elderly man with fever and weight loss
  • Cognitive decline
  • Pancytopenia
  • Lung infiltrates not explained by bronchoscopy and BAL
  • Splenomegaly
  • CSF protein 88
  • Hint of livido

Differential Diagnosis

Is this GCA?

GCA had been the leading diagnosis for 2 months, even in the presence of the negative biopsies, because he had shown some clinical response to steroids.  The family was attached to this diagnosis, as well as the hospital’s team.  In absence of another diagnosis, nobody wanted to stop the steroids.

Is this vasculitis?

In general, this is actually a common situation in the consult service; geriatric patient, nonspecific symptoms, multiple objective abnormalities.  However, they’re not fitting together in a nice neat package or this patient would not have gone without a diagnosis for the past several months. So, herein lies the challenge as a rheumatologist.  Because this case had been stumping the outside hospital, Thomas Jefferson, the medicine team at Hopkins, and other Hopkins consulting services.  Because there were already consults on the chart from both ID and hematology putting vasculitis and GCA at the top of their list.  The persistence of a rheumatologist becomes paramount here, because other diagnoses were not going to be considered until vasculitis and GCA were convincingly ruled out.


Fever in the elderly is a unique entity.  It’s been shown time and time again that older patients have a blunted fever response to infection.  Twenty to thirty percent of geriatric patients will not mount a robust fever in response to even serious infection such as bacterial meningitis. In the fever of unknown origin (FUO) literature, elderly patients are considered patient aged greater than 65 years.  But, in terms of blunted fever response, it’s generally patients in their 80s. Are the older truly colder?  This was first explored in 1991 by Dr. Castle. He wanted to test that the concept of blunted fever response in the elderly was false.  His thought was that there is a significant change in temperature in older patients because the baseline is lower.  Therefore, it’s harder for them to get to 101 degrees, which we classically call a fever.  He took a look at 69 documented infections in 26 elderly patients with an average age of 80 in a VA nursing home in Los Angeles, California.  Half of these patients have what would be considered a blunted fever response; that was defined as a T-max of less than 101 in the presence of a documented infection.  A quarter of those patients had a change in temperature that was greater than 2.5 degrees Fahrenheit, and 90% of the patients had a T-max that was greater than 99, but didn’t hit the 101 peak.  This was the first time that the concept of establishing a basal temperature for geriatric patients, and assessing change in temperature as a fever, was introduced.

In 2005, and then again in 2007, Gomolin specifically looked at the question of basal temperatures and diurnal variation in temperature in older patients.  In 2005, he looked at 100 nursing home patients with an average age of 80.7 years and compared them to 50 community-dwelling geriatric patients. He monitored their temperatures throughout the day as well as their basal temperature.  Most patients had a baseline temperature that was less than what is considered normal, of 98.6 degrees.  The study showed that the 2 groups in general were very similar.  Overall, the oldest patients had the least diurnal variation throughout the day compared to the younger geriatric patients.  This study was highly criticized because it was believed the control group, these 50 community-dwelling elderly patients, was not necessarily the right control group in which to compare the nursing home patients.  Therefore, in 2007 when he repeated the study, he again looked at a group of nursing home residents (with an average age of 82.5), and he compared them to a group of junior high school students with an average age of 14.5.  He compared their temperatures and, once again, the younger patients had more diurnal variation in temperature throughout the day compared to the older patients. The oldest patients with the least variation had a change throughout the day of 0.1 degrees or less, and the junior high school students had a change throughout the day of 0.7 to 1 degree.  These numbers are not very large, but it gets back to the thought of why do older patients have a blunted fever response?  Is it because they just aren’t able to vary their temperature? Maybe fever is not a great way to be looking for infection.

What is a fever in the elderly?  There’s a lot of discussion in the geriatric community about this.  And, there are a lot of recommendations to change the definition of fever in elderly patients to a persistent oral temperature that’s greater than 99 or a change from the baseline temperature of more than 2.5 degrees. There is a lot more research to be done in this area.

Fever of Unknown Origin (FUO)

The current definition of FUO is a temperature of greater than 101 for 3 weeks or more after 3 days of inpatient testing; outpatient testing has been added to the definition as well.  In 1961, Petersdorf and Beeson reported on 100 cases of FUO, and they were the first to organize the causes of FUO into the discrete categories that we use today:

  • infectious disease;
  • rheumatic or inflammatory disease;
  • neoplastic; and
  • miscellaneous.

In 1961, infectious disease was the leading cause of FUO.  Things certainly change over time, and in the 1980s, when Petersdorf looked again at 105 cases of FUO, the top seating of infectious disease had been replaced by malignancy.  And, perhaps that was why, around the same time in the 80s, this study on the Naprosyn test was published in the literature.

Naprosyn Test

The Naprosyn test (1984) touted naprosyn as being a potential tool in distinguishing FUO caused by infectious disease from FUO caused by malignancy.

  • 22 patients with FUO in Oncology Center with documented malignancy or very high suspicion of documented malignancy;
  • Ruled out obvious infection;
  • Treated with naprosyn 250mg, twice a day for 3 days;
  • 14 out of 22 were considered to be responders, temp < 99ºF in 24hrs; all 14 had documented malignancy;
  • 6 out of 22 had no response; ultimately diagnosed with infection, necrotic tumor;
  • 2 out of 2 had a partial response; ultimately diagnosed with connective tissue disorder.

The Naprosyn test is an interesting historical point.  Maybe it’s useful in cancer patients in distinguishing between FUO that’s caused by infection versus their tumor, but it’s difficult to apply to a de novo FUO patient such as the case presented here. Malignancy may have been the number one cause of FUO in the 80s, but its number one status was short-lived.





Petersdorf & Beeson
(1952-1957) USA




Larson et al
(1970-1980) USA




Barbado et al
(1968-1981) Spain




Knockaert et al
(1980-1989) Belgium




Likuni et al
(1982-1992) Japan




De Kleijn et al
(1992-1994) Netherlands




Tabak et al
(1984-2001) Turkey




Zenone et al
(1999-2005) France




All of the above studies looked specifically at patients with FUO, and all had more than 100 patients.  The infectious disease category makes up a very respectable amount of the diagnosis of FUO, and it tends to be the number one cause throughout time.  The percent of diagnoses attributed to malignancy seems to be decreasing over time, and it tends to come in third place.  Percentages of FUOs that are attributed to inflammatory disease have been increasing over time and, in the most recent study, pass infectious disease as being the number one cause of FUO.

FUO in elderly patients

  1. Is the distribution of diagnosis the same when compared to younger patients?
  2. What diagnoses should we worry about?
  3. Where does rheumatic disease fit in?

In 1993, Knockaert specifically looked at this question. He looked at 47 older patients and compared them to 152 younger patients with FUO.  And, if we look at the percent of diagnoses which were attributed to infection in both groups, it was very similar and made up a respectable amount.  There was more tuberculosis in the older patients compared to the younger patients, and this has been shown time and time again. There’s more multisystem disease (considered inflammatory disease) in the older population compared to the younger population, and this makes up a substantial percentage overall in the older population; 12% of these patients had GCA ultimately as their diagnosis.

Although there is more malignancy in the older population compared to the younger population, it makes up, at 12%, overall, much less of the causes of FUO in the older patients (compared to 31% and 25%). This is a trend that’s often seen, which is why we’re much better at making a diagnosis in older patients with FUO than we are in younger patients.

Categories   # of pts (%)
Infection 33 (22.9%)
Neoplasms 14 (9.7%)
Inflammatory disease 38 (26.4%)
Misc 22 (15.3%)
Undiagnosed 37 (25.7%)

In a 1996 study, Zenone identified categories of disease (left). Elderly was defined as greater than 65, and his results were quite similar. The inflammatory diseases that were diagnosed in this study were 11 connective tissue diseases, the vast majority were vasculitis including GCA and PMR.

In general, FUO in the elderly tends to be an atypical presentation of common problem.  Doctors are more likely to find a diagnosis that may be treatable, perhaps not curable, but treatable.  There’s definitely more TB, the malignancies tend to be hematologic. Rheumatic disease makes up a very large percentage of the diagnosis in the older population, particularly GCA and PMR.

Back to the Case

The patient described above fits criteria for FUO and was older than 65. Inflammatory and rheumatic diseases are common causes of FUO in the elderly, and GCA is one of the most common rheumatic causes of FUO in the elderly.

Is this GCA?

The patient fits the demographics for GCA.  At this point, instincts told us that GCA was not the key to this complicated case.  In 2002, Drs. Smetana and Schmerling published a meta-analysis in JAMA.  They looked specifically at which signs and symptoms would predict a positive temporal artery biopsy and, therefore, a diagnosis of temporal arteritis.  They did this by calculating likelihood ratios for the signs and symptoms––a positive likelihood ratio is an increase in the odds of having temporal arteritis as evidenced by a positive biopsy, if the symptom is present.

They investigated a number of symptoms. Symptoms that most likely indicate a positive biopsy and have a diagnosis of temporal arteritis, are diplopia and jaw claudication.  However, other common symptoms that we associate with temporal arteritis and, in fact, the symptoms that are present in our patient, such as anorexia, weight loss, fever, little bit of headache, were not necessarily associated with an increased likelihood of having a positive biopsy. When investigating physical exam findings in a similar format for calculating likelihood ratios, they saw that any abnormality of the temporal artery, whether it was beaded, tender, prominent, enlarged, or the absence of a pulse, indicted an increase in the likelihood that the biopsy would be positive, and a diagnosis of temporal arteritis would be made.  Interestingly, the presence of synovitis made it less likely, probably because the diagnosis lies elsewhere.

Because this patient has no jaw claudication, no diplopia, and normal temporal arteries on exam, GCA was ruled out. There is a negative predicted value that’s associated with temporal artery biopsies of about 90%.  Other tools were needed to look for hidden large-vessel vasculitis. Is there a rule for MRI if vasculitis is suspected and FUO is present?

Wagner et al looked at 67 patients with FUO who were admitted to a rheumatology ward. They looked at 32 patients before, and 35 patients after, the implementation of a standardized workup program for FUO.  The workup program that they used was from the Netherlands FUO Study Group Guidelines and, basically they applied this protocol (example below) to all patients who presented with FUO after 1999 to this institution.


But, in 2002, they added MRI of the aortic arch to their guidelines and to their protocol for workup of FUO.  The amount of diagnoses they were able to make before and after implementation of this protocol increased, but this was not statistically significant.  The increase in the amount of systemic vasculitis that they were able to diagnosis was statistically significant.  Did MRI contribute to this increase in diagnosis of vasculitis? They looked specifically at the patients who were enrolled before and after MRI was added to the protocol, and the prevalence of vasculitis increased from 11% to 42%.  The most prevalent vasculitis diagnosis was large-vessel arteritis.  This evidence led them to conclude that MRI can be a useful tool in an FUO workup, especially if vasculitis is suspected.


The literature states that PET scanning can contribute to 25% to 69% of the final diagnoses.  In 2006, Jaruskova specifically looked at what diagnoses were being made by PET and PET/C:

  • 124 pts w/ FUO
  • 46% (57/124) had a positive PET
  • 89% (51/57) diagnosis based on PET (8 false positives)
  • Out of 51 diagnoses, 17 connective tissue diseases and 10 instances of vasculitis

PET isn’t great in terms of making a diagnosis of vasculitis, but it may be a useful tool.

This patient did not have an MRI, but he did have a PET/CT:


He had diffuse metabolic activity in both lungs, and mild metabolic activity in the left adrenal gland.  Even though he has FUO, and showed a partial response to steroids, he lacked clinical signs and symptoms predictive of a positive temporal artery biopsy.  There were 2 negative temporal artery biopsies, and a PET scan that lit up somewhere else.  These findings indicate it’s very unlikely that this is GCA.

Is this “vasculitis”, or something else?

Stills disease was ruled out because it couldn’t explain a lot of the findings. What about primary CNS vasculitis?  CNS vasculitis definitely could explain the cognitive decline, and it could explain the elevated CSF protein.

What do patients with primary CNS vasculitis look like?

101 patients, diagnosis by angio / biopsy

  • 63% had headaches, 50% altered cognition, 9% fever
  • 22% ESR > 30
  • Median CSF WBC = 5 (0-535)
  • Median CSF protein = 72 (15-1034)
  • Median serum Hg, WBC, platelets

CNS Vasculitis Mimics

  • Systemic vasculitis w/ CNS involvement
  • Creutzfeldt-Jakob Disease
  • Infections: TB, PML
  • Cerebral amyloid angiopathy
  • CNS lymphoma
  • Intravascular lymphoma


Persistence was key at this point. The last recommendation on the chart from the rheumatology service was to get a brain biopsy for this patient. A diagnostic test was performed, but it was done at the bedside.  Four random blind skin biopsies were performed on normal intact skin without any overt findings where the biopsies were taken.


3 out of 4 of the biopsies were positive for intravascular large B-cell lymphoma, and the diagnosis was made by blind skin biopsy.

Intravascular Lymphoma

Intravascular lymphoma (IVL) was first described in 1959. At that time, it was called angioendotheliomatosis proliferans systemica.  It was also called angiotropic lymphoma. It’s a rare variant of extranodal non-Hodgkin’s lymphoma, and it’s classically B-cell in origin.  There are about 2 reports of it being T-cell in origin.  The problem is a proliferation of malignant cells that aree confined to the lumen of small blood vessels or capillaries.  They tend to stick there, which is why there is not a peripheral flow that’s positive.  The signs and symptoms are usually related to which organ is involved.  Most patients have a poor performance status, and almost all patients have B symptoms.  Most of the cases are diagnosed post mortem.

Western versus Asian

It has been suggested that there are some clinical and histopathologic differences between patients diagnosed in Asian versus Western countries. Most of the literature on intravascular lymphoma is case reports; individual case reports, in very small series.  Two of the largest series fit into this geographic distribution. Murase looked at 96 patients, and these were all Japanese patients, and Ferreri looked at 38 patients from Western countries, predominantly in Europe.  When they looked at the phenotypes of these 2 different groups of patients, they saw that there were absolutely some differences.  The Japanese patients tended to have more bone marrow, spleen, liver, and peripheral blood involvement.  Compared to looking at the patients from the Western countries, they definitely had more skin involvement, and they trended to having more CNS involvement.  In fact, Ferreri, in this study, specifically described, for the first time in exclusively cutaneous variant, where the only organ involved was the skin.  The Japanese patients had more cytopenias, and both groups of patients had high LDH, as our patient did.

Is it more than just geography that differentiates these 2 phenotypes? The leading hypothesis is the presence or absence of hemophagocytosis in the biopsy specimen. The classical form of intravascular lymphoma, which is more common, but not exclusive to patients from western countries, has much more CNS and skin involvement and less infiltration of hemolymphoid organs. Hemophagocytosis isn’t seen in these patients.  Then, there’s the hemophagocytosis-related form, which is more common, but not exclusively in Japanese patients.  These patients have more hemolymphoid organ involvement, more hepatic involvement, more cytopenias and they don’t have skin involvement.

Other organ involvement:

  • Neuro à motor, sensory, cognitive, speech
  • Endocrine glands à ADRENAL, hypopit, siADH
  • Lungs à ground glass

Other lab abnormalities:

  • 90% elevated CSF protein
  • 75% elevated ESR
  • Elevated ferritin / LDH

Differential diagnosis:

  • meningitis,
  • Lyme,
  • myelopathy,
  • CJD,
  • primary CNS vasculitis,
  • FUO
  • Common to have dramatic…but transient…response to steroids

How is IVL dagnosed?  In this case, by skin. There’s been a gamut of skin findings that have been reported, everything from nodules to plaques to patches.  The hint of livido was trying to really tell us something, absolutely trying to tell us something about what was going on with this patient.

Random skin biopsy

The idea of a random or a blind skin biopsy really came because of several case reports of intravascular lymphoma being diagnosed, quite frankly, completely by accident.  There is a frequently cited case report of a patient who’s being worked up for Sjögren’s and had a lip biopsy, and IVL was found.  There’s also several autopsy reports where they started sampling organs that appeared to be normal, but started sampling them and looking for intravascular lymphoma, and they found it in unaffected skin. In 2007, a series of 6 patients, very small, who were suspected of having intravascular lymphoma, all received random skin biopsies.  Two of the patients did have skin findings, but they were incredibly subtle, and they not only biopsied the area that was affected, because it was very focal, but they also biopsied unaffected skin.  Each patient had anywhere from 3 to 6 biopsies, so this isn’t just a single biopsy procedure.  But, intravascular lymphoma lesions were found in 23 out of the 26 biopsies.

The important thing is to get the immunohistochemical studies.  In this series, all of the patients had splenomegaly; 5 out of the 6 patients had FUO; 4 out of the 6 had progressive neurologic abnormalities; 5 out of the 6 had thrombocytopenia, and they all had elevated LDHs.  These patients seem to look just like our patient did.

Treatment of IVL

Patients are considered to have disseminated disease when they first present.  We use CHOP or R-CHOP.  Patients, overall, do very poorly.  The 3-year survival is about 30%.  It’s an incredibly poor prognostic sign if the relapse involves the CNS.  High-dose chemo and stem cell transplant may be used in these patients, but the problem is, this population tends to be an older population with a very poor functional and performance status at the time of diagnosis.

Case outcome

This patient received 6 cycles of R-CHOP.  He improved in his mental status quite dramatically.  His family was overwhelmed.  He had repeat random skin biopsies 3 months later, and these showed 0 out of 4 for intravascular lymphoma.


  • FUO in the elderly is a unique entity.
  • Are the older truly colder?  This question is still up for debate.
  • In FUO in older patients, there’s definitely more inflammatory disease compared to younger patients.  There’s more TB, and MRI and PET/CT may help find answers.
  • IVL is a challenging mimic of vasculitis, and there are 2 classic forms; one which involves CNS and the skin. Bone marrow biopsies will not help with diagnosing IVL.  The second form, which is associated with hemophagocytosis, involves the marrow, the spleen and the peripheral blood.
  • Random blind skin biopsies are a noninvasive tool that can be helpful in diagnosing these patients.


Updated: August 16, 2012

Arthritis Center

About Arthritis Center

Founded in 1998, the Arthritis Center at Johns Hopkins is dedicated to providing quality education to patients and healthcare providers alike.