LungRADS 4, 96% Malignancy Probability — Turns Out to Be Tuberculosis

A 61-year-old Hispanic man with more than 30 pack-years of smoking history and a history of recreational drug use was evaluated in the pulmonary clinic for a lung nodule found on low-dose CT. He was born and had lived in New York City all his life, with occasional trips to Puerto Rico, the last one more than a year prior to presentation. He denied exposure to sick contacts. His only symptom was approximately 20 pounds of weight loss over the past year — a finding that would later make more sense in the context of active tuberculosis.

Physical examination was normal: no fever, no lymphadenopathy, no respiratory distress, oxygen saturation 99% on room air, clear bilateral breath sounds. The absence of typical tuberculosis symptoms — productive cough, night sweats, hemoptysis — made the diagnosis far from obvious at presentation.

Figure 1

LDCT (axial): two spiculated RUL nodules — 24 mm and 17.8 mm

CT demonstrating two solid lung nodules with spiculated margins in the right upper lobe, measuring 24 mm and 17.8 mm, with background centrilobular emphysema. No mediastinal or hilar lymphadenopathy was identified. Spiculated margins — radiating linear strands from the nodule into surrounding parenchyma — are the most worrying CT feature for primary lung cancer (likelihood ratio for malignancy approximately 2.8–5.5 in the appropriate clinical context). LungRADS category 4 was assigned with a 40.26% Brock model probability of malignancy. In a tuberculosis-prevalent population or recent immigrant, these nodules would also warrant infectious consideration — but in a lifetime New York City resident, TB was not the first thought.

Given the high clinical suspicion for malignancy, PET-CT was performed, demonstrating metabolic activity with SUVmax of 4 in the right upper lobe. The Herder model — which incorporates PET SUV into the malignancy probability — revised the probability upward to 96%. There were no FDG-avid mediastinal, hilar, or axillary lymph nodes. No prior imaging was available for comparison.

Figure 2

PET-CT (coronal): right upper lobe nodule with SUVmax 4

Coronal FDG PET-CT demonstrating metabolic activity (SUVmax 4) in the right upper lobe nodule. PET SUVmax of 4 with spiculated nodule morphology drove the Herder model malignancy probability to 96% — the diagnostic threshold that justifies immediate tissue sampling rather than surveillance. The critical limitation of PET in tuberculosis-prevalent populations: tuberculosis lesions have mean SUVmax values of 4.2 ± 2.2, directly overlapping with the malignancy threshold, rendering PET unable to reliably distinguish TB from primary lung cancer in endemic regions.

Figure 3

PET-CT (axial): SUVmax 4 — indistinguishable from malignancy by SUV alone

Axial PET-CT confirming right upper lobe FDG uptake (SUVmax 4). Research demonstrates that TB granulomas and primary lung malignancies have significantly overlapping SUV ranges — in endemic regions, the diagnostic accuracy of PET for differentiating benign from malignant solitary pulmonary nodules is substantially limited. An SUVmax of 4 cannot independently confirm or exclude malignancy; tissue sampling is required whenever pre-test probability is high.

Fiberoptic bronchoscopy with endobronchial ultrasound was performed, including needle aspiration of hilar lymph nodes and transbronchial biopsies of the nodules. Visual inspection of the tracheobronchial tree revealed normal mucosa with minimal secretions and no endobronchial lesions. Transbronchial biopsies were obtained from the right upper lobe.

Figure 4

Chest X-ray post-bronchoscopy: right upper lobe opacity

Post-bronchoscopy chest radiograph demonstrating the right upper lobe opacity corresponding to the CT-identified nodules. The chest X-ray appearance — an upper lobe consolidative/nodular opacity in a smoker — is consistent with primary adenocarcinoma, post-obstructive pneumonia, or TB reactivation. Plain film cannot distinguish between these; the CT morphology (spiculation) and PET avidity (SUVmax 4) had dominated the clinical reasoning. Pathology would resolve the ambiguity.

Pathology from transbronchial biopsies revealed non-necrotizing granuloma with bronchial mucosa showing subepithelial chronic inflammation, fibroelastosis, and lung tissue with ill-formed granulomata and multinucleated giant cells. Mycobacteria cultures from the specimen returned positive for Mycobacterium tuberculosis. The 96% malignancy probability was a false positive. The patient was started on standard four-drug RIPE therapy (rifampin, isoniazid, pyrazinamide, ethambutol) and was responding well at follow-up, with serial mycobacterium cultures and pulmonary clinic monitoring.

Figure 5

Transbronchial biopsy: non-necrotizing granuloma with multinucleated giant cells

Pathology demonstrating bronchial mucosa with subepithelial chronic inflammation, fibroelastosis, and lung tissue with ill-formed granulomata and multinucleated giant cells — morphologically consistent with granulomatous infection. Notably, the granulomata are non-necrotizing, which does not exclude TB (TB characteristically produces necrotizing granulomata, but non-necrotizing TB granulomata occur). Special stains alone could not identify the organism; mycobacterial culture — which requires 4–8 weeks — was the confirmatory test. The culture positive for Mycobacterium tuberculosis resolved the diagnosis definitively.

Discussion

This case belongs to the "The Surprising" category that the NLST investigators described when characterizing lung cancer screening's incidental findings. The NLST reported approximately 96% of positive LDCT screenings as false positives; this is one of them — but a clinically meaningful false positive, because the alternative diagnosis (active pulmonary tuberculosis) is itself a serious, transmissible disease that required urgent treatment. The case is not a screening failure; it is screening working as intended, with the caveat that the pathology is infection rather than cancer.

The core clinical problem is FDG-PET's inability to differentiate TB from lung cancer in populations with TB exposure. Research consistently demonstrates that TB granulomata have mean SUVmax values of 4.2 ± 2.2, directly overlapping with primary lung cancer's SUV range. In non-endemic settings, a 24mm spiculated upper lobe nodule with SUVmax 4 and no prior imaging has a genuine 96% malignancy probability by the Herder model. But the Herder model was validated in populations where TB is rare; applying it to patients with prior residence in endemic regions or risk factors for reactivation requires recalibration. The weight loss — 20 pounds over one year — should have explicitly raised TB reactivation on the differential: constitutional weight loss is a classic systemic symptom of TB, more typical of infection than of early-stage lung cancer. The non-necrotizing granuloma on pathology made the differentiation impossible without culture; waiting 4–8 weeks for mycobacterial culture results was the correct and necessary approach rather than proceeding to resection.