Anyone who has endured a multiple-day fever — or cared for someone with one — knows how exhausting it can be. Imagine experiencing nine weeks of a mysterious elevated temperature. Then nine MONTHS of what feels like a personal broken thermostat. And clinicians have little to no explanation besides plans for future tests. That’s the situation that a friend of mine is in and he turned to me for help in crowdsourcing some answers — or at least some new questions to ask.
Here’s the story, in his own words:
At age 65 I have so much to be grateful for. I’ve been married 38 years to a remarkable woman and life partner. We have three affectionate and talented sons who are grown men now. I’ve been immersed in parenting, school counseling, non-profit older adult services, basketball, bread baking, poetry, and woodworking. In 2018 I started an encore career as a chaplain and then counselor in private practice specializing in loss, illness, and life transitions. I’ve been in excellent health for the duration of my life. And then, in March of 2023, my health took a significant turn.
Summary
Since March 2023, I have had several bouts of fevers of unknown origin, some lasting as long as 9 weeks. An extensive infectious disease workup was completed. All testing was negative. Covid and long Covid were ruled out. (As a side note, I did acquire Covid three times prior to March of 2023, all post receiving the vaccine. There were no other infections of significance prior to 2023.) Symptomatically, the medication Prednisone has helped and most recently the medication Colchicine. While early imaging uncovered some inflamed lymph nodes in the stomach, later imaging showed that these were no long inflamed.
Early in the process the medical hypothesis generated was an atypical presentation of Kikuchi-Fujimoto (KF) disease. However, more recently various imaging has shown some additional anomalies including bodies/lesions on the spine and inflamed lymph nodes. (KF disease is less a consideration now.)
Due to being fever-free (since 8/18 with the aid of Colchicine, 1.8 mg daily) and generally feeling well, the plan is to re image (MRI and PET/CT) in October of this year.
Two questions posed to readers:
- Has anyone experienced an auto inflammatory or auto immune condition with fever of unknown origin (FUO) that physicians were unable to label or diagnose?
- Has anyone experienced the above and also discovered through imaging multiple “nodules” (could also be called lesions, bodies, tumors) on the spine?
So, friends, any ideas? Can we point him in a new direction? Please share your questions and thoughts in the comments below. Note that the conversation is never over. If you are reading this months from now, please feel free to post a new question or idea.
Further details in case they are useful:
Recent and Significant Medical History/Testing
February 2023
Right wrist surgery
March/April 2023
Starting in early March of 2023 I experienced fevers of 101 to 103 and fatigue every day for 9 weeks. Many tests were completed for common and rare infectious diseases including blood analysis; imaging by way of XR, CAT, MRI, and PET scans of chest/abdomen/spine; and a bone marrow biopsy.
Test results summary:
3/13 PSA: normal
3/13: Blood: results in general normal (liver, kidneys, sugar, electrolytes); white blood cell count and c-reactive protein (CRP) elevated, which are signs of infection.
3/14 XR Chest: negative
3/17 Blood: Stable results
3/20: CT Chest: A right upper lobe groundglass pulmonary nodule 8 mm, indeterminate, potentially adenocarcinoma spectrum lesion. Also, an indeterminate lesion in L upper kidney measuring 1.3 cm, possible proteinaceous cyst or renal neoplasm.
3/20: CT Abdomen/pelvis: negative
3/22: Blood: Stable results
3/24: Transthoracic Echocardiogram: negative
3/28: PSA: normal
3/28: Blood and blood parasites: stable/normal
4/4: MRI Abdomen:
*1.3 cm lesion involving the posterior upper pole left kidney in keeping with papillary renal neoplasm. No invasion into the subjacent renal sinus fat. No renal vein invasion or metastatic disease.
*Additional renal cortical cystic lesions are consistent with simple cysts. 7 mm right upper pole angiomyolipoma
*Non enhancing cystic lesion of the pancreatic head measuring up to 4 mm, most likely in indolent lesion of epithelial origin
4/14: Blood: Inflammatory markers higher; WBC is a bit lower; moderate anemia
4/14: PET/CT:
*Renal mass to be a cyst—no activity, resolving that issue for now
*Several lymph nodes in the abdomen are only slightly enlarged, but have increased metabolic activity on the scan
*There is suspected FDG avid abdominal adenopathy
*Small pulmonary nodules
4/21: Bone Marrow: The findings are favored to reflect reactive lymphoid aggregates. There is no overt evidence of an intravascular large B cell lymphoma.
May/June 2023
In mid May I started taking Prednisone (60mg/day for 5 days gradually decreasing) and the fevers went away early in the Prednisone tapering. However, when I reached 20 mg/day the fever came back similarly to its previous form. I started Prednisone again at 60mg/day for 5 days (gradually tapering) and this time got to 5 mg and the fevers came back similarly. A third Prednisone trial was started. This time once I reached 20mg/day for 5 days, I went down one mg per week. I was fever free from about 11/23/23 through 3/24/24.
Test results summary:
5/24: Blood: Normal/improved
6/27: Blood: Inflammatory markers (Sed rate, CRP) elevated again, not quite as high as before. White blood cells increased (prednisone has that effect)
6/27: CT Abdomen/pelvis: Nodes slightly smaller, no longer enlarged, scan otherwise negative; went over the scans with the most senior abdominal radiologist. He would have read the most recent scan as “normal’ lymph nodes. Stable 4 mm cyst involving the junction of the pancreatic head and uncinate process
6/29: Blood parasites: negative
July/August 2023
Test results summary:
8/2: Blood: some values off but nothing remarkable
September/October 2023
I had a kidney cryoablation procedure to remove a cyst/tumor (9/5)
Test results summary:
10/9: Blood: some values off but nothing remarkable
10/9: MRI Abdomen: pre-existing lesions: Resolution of all previously noted tumor(s); no new tumor identified since prior study
November/December 2023
I was fever free from about 11/23/23 through 3/24/24; no testing was done
January/February 2024
Test results summary:
1/16: XR Chest: normal
1/22: MRI Abdomen: No measurable viable tumor in kidney area that underwent cryoablation
1/29: CT Chest:
*The CT scan shows no change in the lung nodules. It shows one of the vertebrae in back has fractured and partially compressed. It is the T11 verterbrae. This is new. It is likely a result of loss of bone density from the long course of prednisone
*Stable pulmonary nodules as follows:
–8mm right upper lobe groundglass, possibly adenocarcinoma spectrum lesion
2/2: DXA Bone Density: normal
2/5: MRI Thoracic Spine: compression fracture of the superior endplate of the T11 vertebral body through a lucent lesion that is indeterminate by imaging (similar to earlier findings)
March/April/May 2024
In March of 2024 while in the UK I contracted the RhinoVirus that led to a lung infection. Between March and April I was the sickest I’d ever been. Fevers between 99 and 103 daily and severe fatigue. I also had shooting pain in my legs when standing. (Walking or bending the legs up and down would relieve the pain.) Several cultures for a bacterial infection were negative. After being symptomatic for about two weeks I started Prednisone again and once I got to 30 mg/day (about a week) I started to improve. By mid May I was feeling much better.
Test results summary:
3/18: Blood: Several markers abnormal including Haemoglobin, White Cell Count, Red Blood Cell Count, Neutrophils, Lymphocytes
3/24: Blood: Better then previous but still abnormal including: White Cell Count, Neutrophils, Creatine Kinase, C Reactive Protein
4/3: Blood: Only White Cell Count, Neutrophils, and Monocytes abnormal but much better than previously
4/15: Imaging: (Echo and chest XR) were negative
5/1: Blood: Only abnormal: Mean Cell Haemoglobin
June/July 2024
I came back to the States on 6/4 with the COVID virus. This lead to low grade fevers (99-101) every day for 4 weeks.
Test results summary:
6/21: Blood: some markers off but not remarkable
7/15: MRI Thoracic spine:
*Marrow signal/enhancement: Again as noted a T11 compression deformity of superior endplate without progressive loss of height. There is, however, interval development of multiple new subcentimeter foci of enhancement of numerous vertebral bodies, largest at T3 measuring 5 mm, concerning for neoplastic lesions. No evidence of new compression deformity. New enhancement of the anterior epidural space
7/18: PET/CT:
*Six lymph nodes in the posterior upper abdomen which are top normal in size or minimally enlarged (about 1/2″) and metabolically active. The two nodes seen on PET/CT in April 2023 are no longer metabolically active. No increased activity in the spine, in the area where the MRI noted small areas of enhancement (5 mm and less), and epidural thickening.
*The radiologist who coordinates bone biopsies reviewed your MRI’s and PET/CT scans. He says the process in the spine is unclear. Given the negative PET/CT, he feels a biopsy of spine lesions would be premature.
7/23: Blood: unremarkable with some values off
On 7/5 I saw a Rheumatologist who started me in one .6 mg tab of Colchicine. Fevers abated for about 11 days then came back. I started taking two .6 mg tabs of Colchicine and was fever-free for about two weeks. The fevers came back. On 8/10 I started taking 3 tabs of Colchicine. I was fever-free again starting 8/14 and have been fever-free since.
August 2024
I had a lymph node biopsy procedure on 8/8. A needle was inserted near the right side of spine (stomach area) and 9 cell samples were taken from an inflamed lymph node.
Test results summary:
8/8: lymph node biopsy negative for lymphoma/malignancy
8/27: Blood: Markers improved, the sedimentation rate is normal and the C-reactive protein is lower but still mildly elevated. The alkaline phosphatase is elevated, but only barely given the normal is 126, this elevation could just be lab variability and is rather nonspecific.
Miscellaneous
Family history significant for mother w lymphoma, father w sarcoma, maternal uncle #1 w lung cancer (smoker), and maternal uncle #2 w incurable pelvic tumor.
Image: “Desert Sun,” by Luca Castellazzi on Flickr
I can’t comment on the fevers but when I first developed asthma, I experienced rebounds when my prednisone taper got to 20 or 30 mgs. It eventually became clear that I needed to start a taper at a higher dose and taper much more slowly to knock out the underlying inflammation. I am wondering if something similar could be at work here.
Thanks, Jan, that’s a good call-out to be careful about prednisone tapering.
You asked for people who’ve experienced the same – a “Networker” inquiry, in Rebel Health terms. (Yes?) But my current fire is how people are using LLMs (ChatGPT etc) in cases of difficult diagnoses, and this is sure one of them.
So I used some prompting techniques that I’m learning from the great patient Hugo Campos and from our long-time colleague Gilles Frydman, and then I copy-pasted ALL the details that you put in this post.
[First, a caveat: everyone worries about LLMs hallucinating, and indeed they do. But as Hugo says, “I don’t ask AI to give me the *answer*. I use it to help me think!”]
The result that ChatGPT-4o gave me is here. https://chatgpt.com/share/844e4a41-20ef-46d4-b323-401454662ae4 (It’s long and fascinating!)
______
OK, so here’s some peer-to-peer coaching, first genned up by Hugo, shared with me, and now shared with anyone here. Early tips:
1. Tell it what “role” to play, what kind of person you want it to act like.
LLMs are *designed* to take on whatever personality or voice you say. Go ahead, tell it to respond like a Shakespearean widow! Or, in this case I said, “I want you to consider this case as if you were an expert diagnostician who specializes in cross-specialty cases.” I said that because I want it to focus on the subtle differences between possible diagnoses.
2. When you’re working on a mysterious diagnosis, don’t just say “What could this be?” You’ll get a better answer if you explicitly *ask* it for a differential, aka a differential diagnosis, differential analysis. This is a specific way that doctors are taught to think, and it encourages the thing to use that way of reasoning.
In this case I simply said “What is your differential diagnosis, including the top 3 or 4 possibilities?” (I saw a case today where someone got 15 possible answers – yikes! Too much for my casual interest.)
Give a look at its response at that link above!
_____
p.s. You’ll see two responses. In the first, I hadn’t noticed that you posted the info in two sections, so I’d only pasted in the first batch, and GPT gave me its first impression.
But then I noticed you’d posted a TON of additional info (good for you, good for your friend!), and guess what: with the additional info, the tool adjusted its thinking. I continued the thread: “Some additional new information just arrived. Please update your differential in light of this new history.” and pasted in the new stuff.
It replied: “Given the new and extensive information, including imaging findings, biopsy results, and family history, I would refine the differential diagnosis as follows:”
Of course I have no clue if its response includes the actual cause of the symptoms. As Hugo says, it’s just to help humans think.
And I think we need to do LOTS of teaching people how to use this remarkable new tool.
Dave, I’m grateful for your expansion of the conversation. My friend asked those specific questions in bold because the “peer health advice” frame is what I used to convince him to let me share his story. I sent him examples of other crowdsourcing posts that have been successful, such as:
Case study: Injection tips
https://susannahfox.com/2022/01/06/case-study-injection-tips/
Case study: Trevor’s disease
https://susannahfox.com/2021/10/04/case-study-trevors-disease/
Case study: Brain surgery
https://susannahfox.com/2017/04/25/case-study-brain-surgery/
For some of the above, a fellow human’s experience was essential to helping the target audience. Example: someone preparing themselves emotionally AND learning how to inject themselves safely. Or what it actually FEELS like to to recover from brain surgery.
The Trevor’s disease case study might be a different story. By encouraging my friend (again, anonymous) to share his awesome story of triumph over mystery & pain, we are contributing to the public conversation (and hopefully feeding the algorithm so that more & more people will find the post, as someone did this week).
Those posts were written pre-ChatGPT era. A key ingredient of “digital age chicken soup” (as I started calling information brought to a bedside back in 2013) is now a well-crafted LLM query.
Thank you!
This seems like a good time to share another story of an awesome application of AI to a health mystery.
Longtime readers of this blog may recall Andrew Wilkinson’s story of being diagnosed, after many years of suffering, by someone on Twitter. Here’s the post:
Mystery solved. Again.
https://susannahfox.com/2020/01/07/mystery-solved-again/
I was recently at a dinner party and met a couple who, just chatting about AI, Siri, etc., mentioned that their daughter (age 12) had been suffering for years with digestive issues that no clinicians had been able to figure out. They had traveled all over the country seeking second and third opinions to no avail. Then their kid took out her phone and said, “Hey Siri” and described her symptoms. Siri answered, essentially, “It could be Retrograde Cricopharyngeus Dysfunction.” Friends, that’s what it is. Just like in Andrew Wilkinson’s case the main “tell” was that this 12-year-old could not burp.
As the couple shared the story I said, “Oh! Of course! No Burp syndrome.” I confessed that I was not some incredible diagnostician. I’m a jukebox of patient stories.
Cool, right? All hail Dr. Siri, Dr. ChatGPT (and Dr. Reddit, Dr. Google)!!
First I just wanted say that I feel really bad for your friend. I can imagine this has been really awful for them, and not knowing what’s wrong definitely makes it more stressful.
I am a practicing physician in Boston with an interest in AI and navigation, and I enjoy reviewing challenging cases like this. This isn’t medical advice and is intended just for educational purposes.
It sounds like the level of care has been solid overall. Few things I want to comment on:
1. It’s possible that Covid has more significance in this story than the author might appreciate. Three infections prior to 2023 is a lot. Coming down with Covid for the 4th time in June 2024 after a serious bout with rhinovirus (while immuno-compromised) was a tough blow. That alone could explain why a person might feel weak / not well for months at a time. I doubt long covid is the cause for all of this however.
2. I am questioning the radiologist’s assertion that they shouldn’t biopsy the spine. Not saying they are “wrong”, I’d just really want to make sure this made sense and understand their rationale and get a high quality second opinion. Why not air on the side of biopsying in a case like this? A procedure like this has risk and definitely unpleasant. However, when answers are in short supply, I think it makes sense to be as aggressive as the patient can tolerate.
3. For the biopsy of the lymph nodes around that time, I would seriously consider a pathology based second opinion. This is more common than one might think. They can definitely miss things especially if it’s rare. I would find someone at the best cancer center possible. When I was helping fortune 500 execs with their medical care, I would help them with second opinions from top oncologists and it was quite common for the consulting oncologist to ask that a pathologist from their (elite) institution to do a re-look at the samples.
4. My gut tells me this is a malignancy of some kind, maybe a form of leukemia/lymphoma or metastatic cancer. For a second opinion I would seek out this “rare bird”: a hematologist – oncologist who is a passionate expert in fever of unknown origin (FUO) research. Some people like this topic because they want figure out unsolved medical mysteries. This person probably exists, they just need to be found and traveled to. Worth it. Probably Mayo, Cleveland clinic or some other academic powerhouse. Coming prepared with all the records and forwarding it before hand would be important.
5. It’s interesting but not that surprising colchicine helped like prednisone and was a safer long term solution. It’s an old drug traditionally used for gout. Overtime we’ve realized it’s has multifunctional anti inflammatory properties. It’s used for pericarditis/ myocarditis for example. I’d love to get the perspective of the person who prescribed that drug and why they opted to try it.
Questions I’d want clarity on:
1. What was that renal mass that was removed, what did the pathology show?
2. For the Pancreatic and lung lesions – are we sure they are benign? Were biopsies done?
3. Did you see an infectious disease doctor and rheumatologist?
4. What was the rationale for trying colchicine? Was it a shot in the dark or was there a specific reason?
As for AI, I’ll drop another post below about how I would design a prompt for this as a patient. But just wanted to add another way someone could leverage AI in this case— use it to beef up the clinical summary. Your friend did an excellent job putting this together. You could use chat gpt to take this summary and all of the medical records and use it to make the summary more comprehensive, including not just symptom evolution and test results, but also include what each doctor was thinking and why. Basically, like having a physician drafting up the summary after interviewing the patient and going through everything. Then in a separate chat this new more comprehensive summary could be used to get better, more accurate answers.
Hope this is helpful! Best of luck to your friend. Happy to help if they have any additional questions.
-Dustin Cotliar