P7S Medical Review: Spring 1997, Vol.4, No.1
Clinico - Pathological Conference: A 45-Year-Old Woman With Known AIDS Presents With Sharp Epigastric Pain And Dyspnea

Discussants: Jay Dobkin, M.D., Abbie Knowlton, M.D., Jeffrey Naiman, M.D., and Debra Heller, M.D.

History of Present Illness

The patient is a 45-year-old African-American woman with known AIDS who entered complaining of 4 weeks of fever, 2-3 weeks of sharp epigastric pain and intermittent dyspnea.

The diagnosis of AIDS was established 10 years prior to admission (PTA) when she consulted a physician because of an unspecified rash. Her HIV risk factors included intravenous drug use and unprotected sex with other intravenous drug-using partners. The patient presented with oral candidiasis, which was well controlled with fluconazole. She was also begun on prophylactic oral Bactrim (TMP-SMX) upon initial presentation, which she also tolerated well. The patient reportedly had discontinued IV drug and alcohol use for 10 years PTA, except for a reportedly single, but unspecified, brief alcoholic lapse 4 months ago. At this time, the patient began megesterol at a self administered dose of over 400 mg daily and had gained approximately 20 lbs.

Three months PTA, the patient presented to CPMC complaining of nausea, vomiting and blurred vision for an unspecified time, and was found to be hyperosmolar with a blood glucose level of 1032 mg/dl, pH 7.43, and serum amylase 97 U/dl (nl=0-140 U/dl). Her hyperglycemia responded promptly to fluid and insulin administration, and the megesterol was discontinued. She was discharged within one week with a blood glucose of 100-200 mg/dl. The patient was started on micronase 10mg QD, AZT 200 mg TID and Bactrim DS TID. At discharge, her CD4+ count was 170.

The patient did well initially, but one month PTA developed a fever as high as 103.6 _F and was again admitted. Investigation proved unrevealing with the following serum profile: T Prot 9.5 mg, Alb 4.9 mg, T Bil. 0.6 mg/dl, Bil. 0.1 mg/dl, alk phos 104 U/L, AST/ALT 33 /33 U/L, LDH 31 IU/L, CPK 19 u/L. Sputum for AFB was negative on three occasions, and a PPD was non-reactive. CXR revealed increased bibasilar interstitial lung markings when compared with a film taken two months previously, however, both CVA angles were sharply demarcated at this time. Bronchoscopy revealed no evidence of candida, PCP, or malignant cells. Blood cultures were negative. Echocardiogram revealed no vegetations. CSF exam was within normal limits, with no tumor cells appreciated. Head CT showed no focal intracranial lesions. Abdominal CT revealed diffuse pancreatic calcifications and small periportal and peripancreatic lymph nodes of uncertain significance. A lobular cystic structure was identified in the spleen, and a left adnexal mass was also identified in the pelvis which contained an undetermined amount of free fluid.

Upon discharge, the patient's fever persisted until 2-3 weeks PTA when she developed dyspnea and the onset of sharp, non-radiating, mid-abdominal pain accompanied by chills, constipation and difficulty swallowing. The cough was productive of yellow sputum. The patient denied any chest pain. She reported headaches during this time, and was without visual changes or neck stiffness.

PHYSICAL EXAMINATION

The patient's vital signs were as follows: BP 100/60; P 120; R 23; T 97.9. Upon admission she was in no apparent distress. No skin lesions were appreciated. HEENT examination was significant only for scleral icterus. There was no evidence of pharyngeal exudate or petechiae. Her neck was supple, with shotty suboccipital nodes, but no supraclavicular lymphadenopathy was appreciated. Her lungs were resonant with bibasilar crackles. She was tachycardic as mentioned with a I/VI systolic ejection murmur, best heard along the left sternal border. No cardiac gallop or rub was appreciated. Her abdomen was mildly distended, and tender only to deep palpation, without rebound tenderness or guarding. A non-tender liver edge was felt 2 finger breadths below the costal margin and there was additional evidence of a non-tender, palpable spleen tip. Rectal exam yielded brown, heme positive stool. Gynecological exam was non-contributory. Examination of the patient's extremities revealed no clubbing, cyanosis or edema, and pulses were symmetrical. The patient was awake, alert and oriented x3, with fluent speech, and a non-focal neurological exam.

LABORATORY DATA

WBC 17.1	Urinalysis	Na 134	Ca 9.9	Alk Phos 515
HCT 19.3	SG 1.010	K 4.7	P 5.4	AST 208
Plts 103K	Bil 2+	Cl 101	Uric 7.3	ALT 71
Seg 73	Gluc 0	CO2 22	Chol 132	LDH 960
Bands 11	Micro	BUN 11	T Prot 7.6	CPK 48
Lymph 11	WBC 3-5	Creat 0.6	Alb 2.5	Amylase 46
Mon 5	RBC 2-5	Gluc 83	T Bil 5.9	Lipase 22
PT 20.1			D Bil 3.8	Mg 1.8
PTT 57

pH 7.56, pCO₂ 25, pO₂ 61, O₂ sat 95% Art. Lactate 2.1

HOSPITAL COURSE

On admission, her temperature rose sharply to 104_F. She was transfused after premedication with prednisone, given vitamin K and micronase was discontinued. Sputum was collected for silver stain and for AFB culture and smear. She was given supplemental oxygen by nasal canula and begun on therapeutic Bactrim doses for presumptive PCP. On her third day, Timentin was added, 3.1g Q6 IV. She continued a febrile, dyspneic course though her abdominal pain resolved. Serum glucose remained within normal limits. On hospital day 4 the following laboratory values were obtained: WBC 5.9, Hgb 19.2 with 69 segmented PMNs and 15 bands, electrolytes were within normal limits, and blood sugar 57. PT and PTT remained prolonged, and other tests of liver function remained largely unchanged.

On hospital day 6: INH, Ethambutol, ofloxacin and kanamycin were added. Her dyspnea was such that bronchoscopy could not be done safely and a prolonged PT precluded liver biopsy. She suffered a cardiac arrest on her 7th hospital day and could not be resuscitated.

DISCUSSION

Dr. Knowlton: It is reported that the patient has a history of pancreatitis. Her abdominal film had shown that she had a lot of calcium in her pancreas, is that right?

Dr. Naiman: Yes. The abdominal CT (image #8, without contrast) had shown significant calcification. It was done on August 14th, approximately a week and a half before she died (transaxial abdominal CT shown-1). On this particular slice, the pancreas is diffusely calcified, consistent with chronic pancreatitis. The patient did have a history of alcohol consumption. There is evidence of gallbladder wall thickening, a common finding in AIDS patients. Within the spleen, there is a low attenuation lesion, the appearance of which is not specific. It could be infectious, traumatic or neoplastic. The spleen measured 13 cm. in a caudal-cranial direction and the liver was also enlarged. The patient had a CT one month previously; the liver and spleen had enlarged compared with the prior CT.

Next I would like to show a gallium scan done three weeks before she died. Gallium is an iron analog that deposits in areas of inflammation. The lungs normally have very little uptake; the liver, spleen and colon normally pick it up. This scan was performed 48 hours after the injection. The only significant finding on the planar image is mildly increased uptake diffusely in the lungs. This could be due to infection, any granulomatous disease, and even tumors such as lymphoma, although, lymphoma might be more focal than this diffuse, hazy activity.

SPECT imaging, or cross-sectional imaging of the gallium scan, was done. The spleen did not have a focal absence of radiotracer in the area of the lesion previously seen, therefore the hypodense lesion is unlikely to be a cyst, but maybe an abscess or some other inflammatory lesion.

Figure 1: Transaxial abdominal CT demonstrating significant pancreatic calcification. In addition, there is evidence og gallbladder wall thickening. Both the liver and the spleen are enlarged.	Figure 2: Chest X-ray with vague and patchy opacity at the base of the right lung. The remainder of the lungs are clear. The heart is enlarged.

Finally, the chest x-ray done on admission shows some vague and patchy opacity at the right base (shown-2). The appearance is non-specific, and could be either atelectasis or an infiltrate. The heart was mildly enlarged, and the remainder of the lungs was clear. No lymphadenopathy was seen on this x-ray. There was no evidence of effusion on the frontal view; however, on the lateral view, the major fissure was mildly thickened. This could be due to chronic pleural thickening or a small pleural effusion.

Dr. Dobkin: Is this the typical chest x-ray of miliary TB?

Dr. Naiman: No, but it certainly does not exclude that diagnosis.

Dr. Dobkin: Can you see any evidence consistent with any form of TB?

Dr. Naiman: Tuberculosis can present in almost any way, but there is nothing about this film that would lead you to a diagnosis of TB without supporting clinical information.

Dr. Dobkin: What are the findings that we look for in an AIDS patient in particular?

Dr. Naiman: Most commonly it is lymphadenopathy in an AIDS patient contracting tuberculosis. Cavitation is not as common in AIDS-infected TB patients as it would be in a non-immunocompromised patient, because the AIDS patient is unable to mount the normal inflammatory response, granuloma formation. A true miliary pattern is the most common presentation for miliary TB in an AIDS patient.

Dr. Dobkin: I think it is far from clear here that this was disseminated TB. It was certainly not clear to the physicians managing the patient. One of the chronic and recurring problems in managing TB is that especially when it is extrapulmonary or disseminated it can be a very cryptic process. The literature is replete with cases and series, and anybody who has worked in a TB endemic area has had this experience of making the diagnosis at autopsy. This is not a phenomenon limited to AIDS patients, as it has been particularly described in elderly patients.

One of the key aspects in the management of AIDS patients is the recognition that they are simultaneously at risk for multiple complications. The rule of parsimony in clinical diagnosis went out the window when AIDS came in.

I wanted to say some things about the history itself. What can be made of the statement in the protocol that the patient had AIDS for ten years? The "AIDS" terminology, if it is used precisely, means something that would be very unlikely in this case, so I assume that it is not being used precisely. Having had an AIDS defining diagnosis, which since 1993 includes a T-cell count <200, would be greatly significant. This patient almost certainly did not have "AIDS" for ten years -- that's way beyond the median or even the exceptional survival, so most likely she was known to be HIV-infected for ten years.

Where in the spectrum of HIV disease would this patient lie, and on what basis? And would that impact your differential diagnosis?

The best basis we have now to stage patients prognostically as far as the natural history of the disease is the CD4+ count. It is absolutely critical that it be accurate, because a good deal of management will be based on this value, as will the proper differential diagnosis.

What do you make of the social setting? We hear about a lot of alcohol and IV drug abuse, both personally and in the family. In fact she had a brother who died of AIDS. Is that of any significance? The inference that this setting would put this patient at an increased risk for exposure to TB, and possibly reexposure to TB, is a critical point. The efforts to date to explain differences in natural history by gender or socioeconomic status, however, have not come up with a biologic explanation. As an aside, steroid use, silicosis, cancers and other conditions (e.g. the environment) have been associated with an increased risk of TB reactivation. What's absolutely clear though is that nothing is in the same league as HIV as an accelerant for progression to active disease in a PPD+ or infected individual. Estimates range from 10-100-1000x greater likelihood in the presence of HIV. That and living in New York are more than enough to place her at plenty of TB risk.

The differential diagnosis in this case can be formulated from the following "problem list:"

1) Abdominal pain

2) Dyspnea

3) Dysphagia

4) Abnormal chest x-ray

5) Pelvic mass

6) Increased LFTs:

Alk phos and bilirubin: extracellular/obstructive

AST/ALT: cellular necrosis

7) History of diabetes -- pancreatic insufficiency

8) Blood gases -- tremendous Aa gradient (=60). (NB: Anything >30 in an AIDS patient with PCP is considered severe disease and an

indication for adjunct steroid use.)

9) Pancytopenia -- with a significant leftward shift. (NB: Not ordinarily seen in febrile AIDS patients with AIDS-related diagnoses.)

10) Enlarged spleen on CT.

11) History of alcoholism and pancreatitis.

12) Coagulopathy

So, how do we connect these things? To start from the most general, fevers are very common in AIDS patients, for obvious reasons, and sometimes you can't find an etiology. The patient was sent home on the penultimate admission with a fever, a practice unheard of prior to the AIDS era. Although invasive procedures could have been done to discern etiology (e.g. lymph node biopsy, liver biopsy), often simply waiting for the "usual suspects" to be investigated will reveal the source. We've gotten in the habit, and I think it is defensible, of not tearing patients apart each time there is an unexplained fever, because it is so common. But I would underline the point that this patient was sent home without a diagnosis on the next-to-last admission, and my guess is that what she finally succumbed to was the cause of that process.

Fever could go along with everything else in the differential diagnosis. The dyspnea clearly fits with the A-a gradient, although there is a description on the physical exam that she was comfortable, which is hard to imagine. The most likely explanation for this gradient in a patient with a fever and an abnormal chest x-ray is pneumocystis carinii pneumonia (PCP). She is at risk in particular because of a T-cell count <200. The admitting physicians thought this enough of a concern that they treated her with high-dose Bactrim for PCP, which I think was appropriate. She was also given premedication with prednisone for transfusion. If true, should she have gotten adjunct prednisone for PCP treatment?

PCP has a high mortality rate -- 20-40%. PCP with an A-a gradient like this probably has a 60% mortality rate. The use of adjunctive steroids early in the course has been shown to reduce the mortality rate by as much as half. It is clearly a very important intervention, yet it is obviously also a double-edged sword. You can imagine when this was initially done, no one was comfortable giving steroids to patients that were already severely immunocompromised, but it is lifesaving. The problem is that you need to make sure you're dealing with PCP and not something else. The patient should have had a definitive diagnostic procedure right away, e.g. a bronchoscopy. Expectorated sputum is really not useful. So I would not be surprised to find out that she had pneumocystis and something else along with it.

Coagulopathy, pancytopenia with the left shift and the fever all make me think bacterial sepsis. The elevated A-a gradient can be explained as part of that process. The only thing we don't have is an infectious source or a positive blood culture -- maybe that came back post-mortem. She was treated with an additional antibiotic and could have benefited from additional antibiotic therapy despite her broad spectrum coverage.

What about the dysphagia? On this presentation, apparently no candidiasis was seen. Herpes, CMV, Kaposi's sarcoma, occasionally other tumors could all account for dysphagia. Could that in any way be related to the diagnosis of TB? I can't think of how. You can get TB of the GI tract, typically small bowel and colon. I'm just not aware of a presentation of TB of the esophagus as dysphagia. A lymphoma with widespread dissemination might fit better as a unifying diagnosis.

What about signs of biliary tract obstruction, liver infiltration and splenomegaly? Does that fit in particular with the possibility of TB? This is dramatic in two ways. The physical findings and the radiologic findings are there, and the chemical changes have occurred in the last month. All of these chemistries were normal on the previous admission. It could relate to extrapulmonary or disseminated TB, which commonly involve the viscera. Liver biopsies have probably the highest diagnostic yield in extrapulmonary TB, but there is a much bigger differential, especially in AIDS patients.

Along with lymphoma, disseminated MAC is the most common cause of this presentation, but her T-cell count is not low enough (>50), if it is accurate. Therefore, it is critical to know the true count.

Disseminated fungal infections of several sorts, disseminated yeast infections with cryptococcus, histoplasmosis, and in the right setting, coccidiodomycosis, can all do this. In fact, if you wanted a single AIDS-related diagnosis to explain this whole presentation with a septic appearance, multi-system disease, hepatosplenomegaly, disseminated histoplasmosis would probably fit the best.

What's the one problem? If we assume she's a New-York born and raised African American, she hasn't been exposed to histoplasmosis. The cases we see are essentially all reactivation cases of people from endemic areas: the upper midwest, the Mississippi/Ohio River valley, large parts of South America and, significantly, almost the entire Caribbean. So the disseminated histoplasmosis that we have seen here is typically in people from Central and South America. It is not particularly limited to the lowest T-cell counts.

CMV, for the same reasons, would not fit with the T-cell count of 170. It would more likely explain the transaminase elevations, but not the organomegaly. It would be very hard to account for the pulmonary picture, as CMV infected patients do not typically have this septic presentation. The stereotypic CMV syndromes in AIDS patients include retinitis, GI disease, ulcers and inflammatory processes, and increasingly, a variety of CNS syndromes.

In the end, I'm coming up with TB for one strong reason: not that there is anything about this protocol that suggests it, but that she was clearly at high risk and it is a very, very treatable condition, even in AIDS patients. People not only survive TB, they appear to be cured of it. It is really a shame to miss the diagnosis, yet as I said at the beginning, it is often missed. What you have to do as a clinician, if you haven't come up with another explanation after a certain point, begin empiric antituberculosis therapy. That is what they did, but obviously it was way too late. You cannot wait for a proven diagnosis in these patients. Even with new technology, like PCR for TB detection, there is no foolproof way to prove or rule out the diagnosis, especially of disseminated TB.

Dr. Heller: The gross pathological findings were actually somewhat limited to the abdomen. The chest, except for the fact that the lungs were consolidated, had no focal lesions and no miliary TB grossly. In the abdomen, the patient had about 500cc of ascitic fluid. The liver and the spleen were both enlarged, and the mass in the pelvis was a small fibroid on the uterus. The pancreas was diffusely calcified and fibrotic.

In terms of gross findings once the organs were internally inspected, the lungs were consolidated diffusely, as if there was a pneumonic type of process, but there were no focal lesions. The only focal lesion was in the spleen, and what you see here is an abscess cavity, 3 cm in its largest dimension.

We were somewhat surprised histologically, because the picture in the spleen fits miliary TB. There were diffuse microscopic granulomas throughout most of the organs in the body. Normally, in miliary TB we see little punctate lesions covering most organs grossly. This was not seen here. The nodular granulomatous inflammation does not have central caseation as you'd expect, but this is not uncommon in AIDS patients.

This is the patient's bone marrow, which is somewhat hypercellular; usually it is about half fat and half cells in this patient's age group. The bone marrow also had granulomatous infiltration (shown-3).

The kidney histologically showed some confluence of granulomas, with some evidence of necrosis (shown-4). But this was not seen grossly either. The lesion in the spleen was also granulomatous and proved to be TB.

The paratracheal and intrabdominal lymph nodes were also involved.

The patient's liver had some fatty infiltration of the remaining parenchyma, consistent with the history of alcoholism. The fibrous nodules of cirrhosis were not appreciated, but there is some fibrosis in portal areas. Again, multiple granulomata were seen.

The acid fast stain at low power demonstrates the normal liver parenchyma separated from these granulomata (shown-5). We did these stains on many of the patient's organs, despite the fact that this is not particularly sensitive for TB. A culture is much more sensitive for making this diagnosis. In another area of the patient's liver, fatty infiltration was seen, as was a lymphoid aggregate in a portal area, which is highly suggestive of hepatitis C.

Regarding the patient's pancreas, very little parenchyma remained, as it was largely replaced by densely fibrotic tissue.

There was a pre-mortem blood culture, taken a few days prior to death, which came back a month later positive for M.tb. We didn't find any other infectious diseases.

Tuberculous granulomata were also evident histiologically within the lungs (shown-7).

Figure 3: The histopathelogical bone marrow section demonstrating granulomatous infiltration.	Figure 4: A histo pathological kidney section with some confluence of the granulomas and some evidence of necrosis.
Figure 5: An acid fast stain of the liver demonstrates the normal liver parenchyma separated from the granulomata. Rare acid-fast bacilli are seen.	Figure 6: A histopathological section of the lungs demonstrating tuberculous granulomata.

Dr. Naiman: The following is a list concerning the prevalence of different chest x-ray findings in patients with miliary TB. I think the most important thing to note is that having a miliary pattern is very common, but its absence is not uncommon in miliary TB. There are many nonspecific patterns, including interstitial and interstitial/alveolar. Cavitations, if you see them, may suggest tuberculosis, but they're not common at all.

Dr. Dobkin: At this point, the distinction between extrapulmonary TB, disseminated TB and miliary TB must be emphasized. Miliary TB is by definition disseminated TB, but in my experience, only a tiny minority of cases of disseminated TB have a miliary pattern on the chest x-ray. So the chest x-ray and the pulmonary work-up is only a minor part of the approach to making a diagnosis of disseminated TB. If suspected, what really must be done, as in this case, is obtain tissue for histological study: bone marrow, liver, lymph nodes, even blood. An aggressive workup is necessary -- no one investigation has a particularly good yield. Concerning this case, I would not conclude miliary TB since in my lexicon that requires a miliary pattern on the chest x-ray. Disseminated TB with multi-organ involvement most of the time does not have a miliary pattern, as seen in this case.

I think cryptococcus would have been a strong consideration to treat. It certainly wasn't ruled out in this case, and the absence of cryptococcal antigen in the spinal fluid cannot be assumed to rule out visceral disease.

The chest x-ray appearance of TB in AIDS patients is strongly conditioned by the CD4+ count, as is the risk of acute infection and reinfection. In this case, because of the time course, this may have been a recent infection, despite her relatively elevated CD4+ count for this process. Clearly she had a widely disseminated infection.

The fact that this may have been acute, or simply the fact that she was desperately ill in New York City in 1995, also raise the possibility of drug-resistant disease. I am not sure what the regimen that was finally chosen was based on, but it was not a good regimen for the kind of multi-drug resistant TB that we have been dealing with. So there really is a definite role for casting a wide net and if you cannot make the diagnosis and there is still a possibility, you begin empiric therapy.

Dr. Knowlton: The risk of the drugs, we feel, is outweighed by the possible risk of the disease.

Dr. Dobkin: By the same token, it cannot be done casually. It is really not appropriate to throw TB drugs at a patient with no work-up, because what if they get better and you do not have a diagnosis? Then you get into drug toxicity. It is a big challenge, but has always been a big challenge. It is really not changed a lot, except that there is more of it because of AIDS. There was a series published years ago on TB cases diagnosed at autopsy, and it is the same process over and over: they thought of it, but they were waiting for something to come back positive, and the patient ends up in the morgue.

This patient could have been cured, and could have, presumably, lived for five years or more, had she been treated for TB in a timely fashion.