Absolute certainty in diagnosis is unattainable, no matter how much information we gather, how many observations we make, or how many tests we perform. Our task is not to attain certainty, but rather to reduce the level of diagnostic uncertainty enough to make optimal therapeutic decisions.1
Proponents of a clinical diagnosis of stage I sarcoidosis (SIS)—isolated bilateral hilar adenopathy (BHL)±right paratracheal adenopathy, erythema nodosum, or uveitis—advance the view that its features are so distinctive that confusion with an alternative diagnosis requiring intervention (AD) is exceedingly unlikely. Moreover, because its course is typically brief, benign, and self-limited, histologic confirmation is not required; ongoing clinical follow-up suffices. Proponents of tissue confirmation advance the argument that delayed diagnosis of an AD simulating SIS, for example, lymphoma, may lead to irretrievable harm.
In 1973, Winterbauer et al2 undertook the seminal, systematic study of BHL to resolve the controversy surrounding the necessity of tissue confirmation in presumptive SIS. Among 100 patients with BHL of various etiologies, including lymphoma and solid neoplasia, sarcoidosis was found to be the cause in all 30 who were asymptomatic and had a negative physical examination. The authors concluded that a confident a priori diagnosis of sarcoidosis could be made in patients meeting these criteria, and that biopsy confirmation was unnecessary.
To augment the robustness of this conclusion, we3 computed the absolute incidence (incidence of each condition×proportion presenting with isolated BHL) of ADs—Hodgkin disease, non-Hodgkin lymphoma, and tuberculosis (TB)—versus the absolute incidence of SIS (incidence of sarcoidosis×proportion presenting with isolated BHL). We estimated that if 33,000 persons with presumptive SIS underwent mediastinoscopy, sarcoidosis (or, rarely, a disorder not requiring intervention) would be confirmed in 32,982 (99.95%), and, at most, 9 persons would be diagnosed with Hodgkin disease, 1 with non-Hodgkin lymphoma, and 8 with TB. Our estimate of the absolute incidence of ADs would have had to be off by a factor of >100 to nullify this extension of Winterbauer’s findings. York University, at the behest of the British National Health Service, vetted this paper, and agreed with its findings. The National Health Service adopted a policy of accepting a clinical diagnosis of SIS.
Brute empiric quantification of the incidence of ADs simulating SIS reinforces this assessment: Assuming a combined population of 1×109 for Western Europe, England, Canada, Australia, and the United States, and a sarcoidosis incidence of 3×10−5, there would have been 120×104 cases in the 40 years since Winterbauer’s seminal paper. Assuming half are stage I, there would be 60×104=600,000 cases susceptible to simulation by an AD. If as many as 5×10−4 ADs presented with isolated, asymptomatic BHL, there would have been 300 such instances. To date, not a single validated instance has been published. In brief, one would have to subject 10,000 persons with SIS to an invasive procedure (with its accompanying costs and morbidity) to identify, at most, 5 persons with an AD.
Justification of histologic confirmation of SIS requires that proponents furnish: (1) a case series demonstrating a substantial number of ADs simulating SIS; and (2) evidence that the diagnostic delay imposed by watchful follow-up (in place of histologic confirmation) imposed substantial harm. (It should be recalled that an AD will become evident with routine follow-up of SIS with little harm as a consequence of any diagnostic delay.3)
Primary TB: Histologic verification may appear justified in regions with a high prevalence of TB. Assuming that the UK TB-incidence figure (14×10−5) is representative of the Western population (of ∼1 billion), there would be almost 6 million new cases of TB in the 40 years since the seminal analysis of Winterbauer. Not a single instance of asymptomatic tuberculous BHL simulating SIS has been reported. Thus, hundreds of thousands of invasive procedures in persons with asymptomatic BHL would be required to identify, at most, 1 case due to TB. Moreover, primary TB is self-limited in >90%; and progressive primary TB would become evident with routine follow-up. On the basis of these considerations, the regional prevalence of TB is irrelevant to a decision to obtain tissue verification.
Silica and AIDS: Instances of asymptomatic BHL due to silica exposure and AIDs have been reported. Where indicated, an inquiry into silica exposure will reliably and efficiently exclude the former, and a serological test will exclude the latter.
In contrast, Boujaoude et al4 justify their policy of tissue confirmation of SIS by citing an authority: “…several major pulmonary organizations published a combined statement on sarcoidosis that included the recommendation that routine histological confirmation of diagnosis in BHL be obtained regardless of the presence or absence of symptoms.” This statement5 contains no such recommendation; its authors supplied this view of histologically unconfirmed SIS: “Some patients refuse biopsy…. Clinical and/or radiological features alone may be diagnostic for patients with stage I (reliability, 98%)….”
Boujaoude and colleagues, citing an article by Carr et al,6 go on to state “The validity of this approach [clinical diagnosis] was brought into question in 1990 by an epidemiological analysis that calculated that 30% of cases of BHL will not be due to sarcoidosis and that lymphoma [21%] will be the next most common diagnosis.” Carr and colleagues erred in estimating the incidence of ADs on the basis of the proportion of all cases (of lymphoma, neoplasia, TB, and some rarities) with extant (inclusive) BHL—that is, not isolated (exclusive) BHL. Were lymphoma the etiology of 21% of isolated, asymptomatic BHL, 126,000 instances (0.21×600,000) would have by now appeared; however, not one has been reported.
The main flaw in the paper by Boujaoude and colleagues lies in its conflation of isolated, asymptomatic BHL with mediastinal lymphadenopathy, followed by the imputation that the conflated findings apply to each individual condition. Mediastinal lymphadenopathy is a frequent feature of Hodgkin and non-Hodgkin lymphoma. Dr Boujaoude (written communication, January 10, 2013) confirmed that none of the lymphoma cases reported in his series presented with isolated, asymptomatic BHL.
I know of no condition in which the disparity between the costs and morbidity of securing a histologic confirmation by invasive means so greatly outweigh its utility. The following explanations, none of which are valid justifications, appear to be likely contributors:
1. Excluding cancer is intuitively considered a categorical (ie, unconditional) good. This intuition may, for example, contribute to the persistence of prostate-specific antigen screening in octogenarians and the performance of Pap tests in posthysterectomy patients.
2. A radiographic interpretation of isolated BHL incorporating TB, fungus disease, and lymphoma (along with sarcoidosis) in the differential may influence the decision to proceed with tissue confirmation, if only to forestall the possibility of litigation for failure to diagnose.
3. Availability of endoscopic, ultrasound-guided, transbronchial biopsy, a more sensitive procedure than unguided transbronchial biopsy for the diagnosis of SIS and a far less invasive procedure than mediastinoscopy, may lower the threshold for securing tissue confirmation.
4. The expectation of referring providers for confirmation may not be satisfied by an assertion by the consulting pulmonologist that he can make a confident diagnosis without tissue confirmation. This view might be shared by the subject who may require further assurance.
5. On the part of the provider: (1) lack of understanding that the pretest probability of SIS, suitably defined, exceeds 99.95%; (2) pecuniary advantage; (3) it is difficult to judge how great a part, if any, is attributable to traditional thinking, what Thomas Sowell, in another context, referred to as “The vision of the anointed; the irrelevance of evidence.”