CRITICAL REVIEW Miliary Tuberculosis: Diagnostic Methods with Emphasis on the Chest Roentgenogram* Herbert \\'. Berger, M.D., F.C.C.P.,o 0 and Teresita G. Samortill, M.D. t
The initial chest roentgenogram was reported to be normal in 12 of 14 episodes of milial)' tuberculosis. On review, miliar)' lesions were seen on ten films. The four normal roentgenograms exhibited milial)' infiltrations after periods of se\'en days to one month. Liver and bone marrow biopsies were more reliahle diagnostic methods than sputum cultures. The initial chest roentgenograms may frequentl)" be interpreted as normal in miliar)' tuherculosis, particularl)' when the radiolo~ist does not ha\'e detailed clinical information, but with careful review of initial and follow-up films miliary lesions will become apparent in almost all cases. The earl)' roentgenographic detection of milial)' tuberculosis depends upon clinical suspicion, technically adequate films, and close cooperation between clinician and radiologist.
There are conflicting opinions regarding the frequency of chest roentgenographic abnonnalities in miliary tuberculosis. I-Ii 'lost physicians seriously consider this diagnosis only when charncteristic lesions are seen in the lung fields, while others have stated that roentgenograms are often normal even in well established cases.I--l During the past five years. we have observed 1.3 patients with proved miliary tuberculosis. This report re\'iews the frequency of abnormal chest roentgen findings in these cases and correlates them with their pertinent clinical feahlres.
massin' Iwmatogenol1s dissemination of tubercle bacilli to the lungs and other orl!,lIlS producing generalized systemic illness, The diagnosis of miliary tuherculosis was confirmf'd in evpry patient by culture for acid-fast organisms, biopsy. or autopsy, Biopsy fincJings wcre considered compatible with tuberculosis if eitlwr c(1st'ating or Jlollcawatill.1! epithelioid granulomas wert' foulld eH'1l whell ac:id-fast bacilli could Jlot be de1l1ollstrated. \\'l' examined eal'h patiellt's chest roentgt'nograms at the time of hospitalization. amI reviewed and analyzed the entire series of caSt'S subsequelltly, The chest films WPft' originally reported by hospital attending radiolo\!ists who often did not haH' detailed dink-al informatiou. alltl then were ft'\'iewed hy them with the authors for possiblt' ft'\'isiOIl of illterpretatioll. The roelltgcllograms Wt'n' cOllsiderec:l abllormal if miliary infiltrations 1-2 m1l1 in diameter Werf' noted, Clinical data were gathered from hospital charts and from our own consultatiolls aud Ilotes. as well as from autopsy protocols,
~IETHODs Froll1 .lilly. H)(i4 throll.l!h Jllne. 1969 a diagnosis of miliary tuberculosis was estahlished in 1:3 patients at the \Iount Sinai Hospital S('T\'ices-City Hospital Cenh,'r at Elmhurst. One patient had a similar episode in 1962 so that we report a total of 14 episodes, \\"e followed the usual definition of miliary tuherclllosis-
RESl'LTS
Clillical Features Information concerning age. sC'x. tuherculin alwrgy. erythrocyte sedimentation rate (ESR), cultures for hlberck bacilli. biopsies. and duration of fever after starting antituberculosis therapy is summarized in Table 1. Except for fe\'er. which was present in every patient. the history and physical examinations revealed no uniform abnormalities. As a group, the patients were ill from one to eight weeks prior to hospitalization. \\'e did not find choroidal tubercles,
° From the Dt'parhnent of \fedic:ine, \follnt Sinai Schexll of \Iedicine of The City University of :\ew York and \Iollnt Sinai Hospital Servit:es-City H;lspital Center at Elmhurst, Elmhurst. :\cw York. ° °Chief. Pulmonary Disease SeT\'ice. \Iount Sinai Hospital Services-City Hospital Center at Elmhurst; .\ssociah,· Professor of Clinieal \Iedicine, \fount Sinai School of \Iedicine, Tht' City lTni\'t:'rsitv of :\ew York, tResident in \Iedicine, \'Iount Sinai Hospital Senices-Citv Hospital Center at Elmhurst. .
586
587
MILIARY TUBERCULOSIS TaM,· I---#.'Ii"i..", F.."t",... ill 13 P"ti..",. with ,tliIi,,,,'
T"I"·,.."I,,.i.
,\\.:
.. <:~.)
yr.....
>:,O,\T', :\\,'n
II
\\'lIf1Wfl
TIII ...r"IIIi" ,\rwr\.:v Pn'~f'llt
.\h·"·II' E:'H
"r
<20111111 hI"
:1
>211
i
IIIIIl
('lilt lln':-,
:--:rlH'ar...
:-'ptlllllll
,., III
Bo"O' :\I"rrow l'rinO' ( ':,F
:! :!
Liv"r
2 ,.,
I :1 11:1
'ai ,II·
'\ >II
Bi11p,"It':-' 1)II:"i11\".· Tl1fM·r,olJln"'l:",
Liv"r !lo"" :\I"rrow
:'",,1,,1,,' "0010' :\\ollt" ('I,'O'r
!luration F"\"'r ,\fr"r TI... r:ql\'
>2101a\" < 21 oIa\',
IniTial l)j;q!f1l1... i...
:\\ ilia!'." T,aI.'r"III. "i,
Ilt I... r
, III
Ftr;nIE
I, Clll"t rDl'ntl-';enogram showing miliary infiltrations,
I )"" I ",
From clinical or autopsy t'\'idt'nce. the sOllrct' of spn'ad was thought to he a caseous focus in tht' lungs in fi\'(' patit'uts. iu the kidlll'\'s in two. in the peritOlH'um in ont'o and \\'as unknown in thl' otlwr fi\'(' patit'nts. TIll' diagnosis of miliary tulwrculosis was suspedt'd at tIll' time of admission to the hospital in fom patients, In d('n'n episod('s thl' diagnosis \\'as ,'stahlislll'd during life. and in tlm'e instancI's c1is(1)\I'n'd lIIW\lwdt'dly at autopsy,
I\t'mato~t'nolls
Ratl;olo~ir' Fca/rm'v
Smprisingly the initial dwst nwntgl'nogram was n'portt'd to he norm,1I in l~ of the l-t episodes, ,\ltt'r careful n'dl'\\' of thl'st' films. faint miliary Il'sions WNe found in tl·n. while fom were still 11111sidef('d normal (Fig I. ~). In each of these fom patients with no ahnormalities on tlwir initial dwst fillll. miliary infiltrations could hI' seen SI'\'I'n (Ia~'s to on,' month later. Two other patients dl'monstratt'd rO('ntgenographie progression o\'('r spans as little as four and It'n dan. The distrihution of the I('sions and rate of dl'aring aft('r tl)(' initiation of antitull('reulosis drugs are notl'd in Tahle 2.
CHEST, VOL. 58, NO.6, DECEMBER 1970
FIGl'Rt: 2, Film four days later showinc an inert'a'" in tl ... nlllllhn of miliary I.·,ion,.
588
BERGER AND SAMORTIN
Table 2-Ra,liolo«ic Feal"re. ill 13 Pa,iell'. wi,h ,Uiliar,- Tuberculo.i. RClf'lIt/!PII :-\tatl1:Initial X-ra~' UPpOI·t :'\ol'ln:11 .\hlloruml
12 2
lllitial X-ray rp\'ic'\wd :'\orlllal ,\IHlormul
-l
10
X-ray c'on\'pr:-oioll from :'\ornwl to .\bnorlllal
-l
6
X-ray prcl/!rps''''iclIl Di:-:trillution of Ip:-oion:-: l'niform l'pppr lun/! J.,owpr 11In~ X-ray (·Iearin/! first notpd aftpr
thprap~'
1 mCI 2 mo
:J
-I
1
-I
1110
:'\onc'
I DISC:USSIO:,\
The diagnosis of milian' tuberculosis is to be considered in any patient' with fever of undetermined origin. Although former1\- considered a disease mainly of infancy, miliarv tuherculosis is now encounted with increasing frequency in adults. Ii A history of pn'\'ious pulmonary or extrapulmonary tuherculosis, particularly if never treated with isoniazid, should alert one to the possihility of reactivation and hematogenous spread, as was found in three of our patients. Although not observed in our series, choroidal tuhercles are highly indicative of disseminated disease and ha\'e heen described in as many as 60 percent of patients with miliary tuberculosis. ;;-1 II A careful search for them should be made after fully dilating the pupils. On occasion, patients may have temporary tuherculin anergy and low erythroeyte sedimentation rates as found in this series. Bacteriologic or histologic confirmation of tuherculosis is crucial to making the diagnosis. \Ve feel that a diagnostic trial of antituberculosis therapy is unwist, because of possible coincidental defervescence in nontuherculous patients. The fair1\- frequent occurren('{~ of prolonged fever after st~rting therapy in tuberculous patients can also he mislead~ ing. Ii. 11 Four of eight patients in our series had fever mon> than 21 deWS after antituherculosi~ therapy was lwgun. . Our initial approach in patients with a suspected diagl!osis of miliary tuherculosis is to obtain smears and cultures of sputum and of urilw or ('('rehrospinal fluid wh('re\'er appropriate. If tulwrde hacilli
cannot be demonstrated on smear, we promptly perform liver and hone marrow hiopsies because of the infrequency of positive sputum cultures for acid· fast hacilli and the long incuhation time required. In the ahsence of a coexisting pm'umonic focus, only three sputum cultUft:·s demonstrated tubercle bacilii in our series. In two of these instances, the miliary infiltrations were exceptionally numerous, and sputum smears revealed acid-fast hacilli; only two colonies were cultured in the third patient. Biehl'; reported positive sputum and! or gastric cultures in 61 percent of 49 patients with miliarv tuberculosis. Our experience confirms the usef~lness of liver and hone marrow hiopsies. 2 .... 111 All five liver hiopsies performed and four of five hone marrow biopsies were helpful in making the diagnosis of miliary tuberculosis. Even when acid-fast hacilli are not seen on the sections we feel that the presence of eith(>r caseating or non-caseating epithelioid granulomas in a patient with a compatihle clinical picture warrants a presumptin' diagnosis of tuberculosis. Culture of Ih'er and hone marrow specimens occasionally yields organisms and thus enhances the value of the hiopsy proeedures wlwn the histologic findings are negative, as found in one of our patients. Cultures are also helpful in estahlishing the diagnosis in cases with epithelioid granulomas, hut no demonstrated acid-fast bacilli. Recentlv we have had fruitful results hy directly innocuh;ting bone marrow and pleura into liquid cultun' media, as descrihed by Scharer and ~kClement.l:! \\'e hope that this tedlllique will be diagnostic, particularly in liver biopsy culhlfes, which were all sterile in our patients. If both liver and bone marrow hiopsies are negative, a needle or open lung biopsy may prove diagnostic. I:: In the present series, the diagnosis of miliaf\" tuherculosis was made on admission to the hospit<;l in only four patients and was unexpectedly made at autopsy in three patients. Chapman and \Vhorton I" and Treip and ~Ieyers 1:-, reported that 2.5 pert'{>nt of their autopsy cases of milian' tuherculosis had been diagnosed prior to death. The undiagnosed cases were im'ariably thought to have normal ch('st roentgenograms. ~liliaf)' infiltrations were initially reported on the admission chest roentgenograms in only two of the 14 episodes in our series. Aft('r review of these films they were found on ten films, hut were still not seen on four. Steiner I found abnormal roentgenograms in 22 of 46 patients. Emery and Lorher:! 18 of .52, Bottiger et aP zero of five. and Heinle et al" three of nine patients. On the other hand, Lewison et aI" found 75 of 96 films ahnormal, and Biehl li 6.'3 of
68.
CHEST, VOL. 58, NO.6, DECEMBER 1970
589
MILIARY TUBERCULOSIS
All four roentgenograms which initially had a normal appearance exhibited miliary infiltrations after seven days to one month. In an additional two patients, including one under drug treatment increasing involvement was found on subsequent films. I t has heen stated that miliary lesions may not become visible until two and one-half to six weeks following dissemination to the lungs. 1 .;.1 'j Based on our experience it is clear that occasional patients with miliary tuberculosis may han' nonnal initial chest roentgenograms, but that after careful review of both initial and follow-up films, miliary infiltrations can he recognized in almost all if roentgenograms are ohtained every four to seven days. The individual miliary l<.~sions were uniformly distributed throughout the lungs in three patients, more numerous in the upper lung fields in three, and more num<.~rous in the lower lung fields in eight. Habin 1 Ii states that the lesions are larger and more numerous in the upper lobes in contrast with our ohservations. Our own experience. as well as that recorded in the literature. demonstrates that chest roentgenographic recognition of miliary infiltrations may be difficult particularly in early cases. The lesions are hest delineated on slightly underpenetrated films when one scrutinizes small areas of lung hetween the rihs. Serial films must be examined in order to recognize dev(~loping lesions. \Ve have found it us(~ful to compare the chest roentgenograms of suspected cases simultaneously with those of patients with proved miliary tuherculosis. The radiologist must he made aware of the c1ink'al infonnation if lw is to recognize faint lesions. The identification of lesions is enham.'ed if the clinician and radiologist review the films together. Steilwr 1 feels that there is a good correlation between the radiologic recognition of miliary lesions and their histologic structure. In his series, abnormal chest roentgenograms were found only in cases with caseation and collagen formation. Emery and Lorber:! agree with earlier authors that number and size of the miliary lesions are the major detenninants in radiologic recognition. Hoentgenographic clearing of nodules was noted in three of our nine patients within one month after starting antituberculosis therapy and clearing began as early as nine days in one patient. Four additional pa tients demonstrated improvement after two
CHEST, VOL. 58, NO.6, DECEMBER 1970
months. In one instance, no radiologic change was seen even after 16 months, but this patient had had previous exposure to silica dust.
REF EREXCES Steiner PE: The histopathological hasis for the x-ray diagnosahility of pulmonary miliary hlherculosis. Amer Re\' Tuherc 36:692, 1937 2 Emery J L, Lorher J: Radiological and pathological correlation of miliary hlherculosis of lungs in children with special reference to choroidal hlherdes. Brit \Ied J 2:702, IH50 3 Bottiger LE. l'\ordenstam HH, Wester po: Dissl'minated tuberculosis as a cause of fever of ohscure origin. Lancd 1: 19, 1962 4 Ht'inle EW, Jr, Jensen \VN, Westennan \IP: Diagnostic usefulness of marrow biopsy in disseminated hlherculosis. Amer Rev Resp Dis 91 :701,1965 5 Lewison \1, Freilich EB, Ragins OB: A eorrelation of clinical diagnoses and pathological findings with spt'l'ial reference to hlherculosis an analysis of autopsy findings in 893 cases. Amer Re\' Tuherc 24: 152, £931 6 Biehl JP: \liliary hlherculosis a review of sixty-pight adult patients admitted to a municipal general hospital. :\1I1er He\' Tullt'rc 77: 60.5, 19.58 7 Illingworth RS, Wright T: Tubercles of the choroid. Brit \Ied J 2::36,5, 1948 8 \Ia'isaro D, Katz S, Sachs \1: Choroidal hlherdes a clue to hematogenous hllx·rculosis. Ann Intern \Ied 60:2.'31, 1964 9 Ol~lzahal F, Jr: Choroidal hllx'rdes a neglected sign. J:\\IA 200:374, 1967 lO Emery .IL, Gihhs 1\;\1: \liliary tuberculosis of the hone marrow with particular reft'rence to the possihility of diagnostic a'ipiration hiopsy. Brit \Ied J 2:842, 19.'54 11 Berger H\V, Ro~enhaum I: Prolonged fever in patients treated for hlherculosis. Amer Rev Resp Dis 97: 140, 1968 12 Scharpr L, \IcClenwnt JH: Isolation of hlherde hacilli . from needle hiopsy specimens of parietal pleura. Amer Hev Resp Dis 97 :466. 1968 13 Adamson JS, Jr, Bates JH: Percutaneous needle hiopsy of the lung. Arch Intern \Ied 119: 164, 1967 14 Chapman CB, Whorton C\I: Acute generalized miliary hlherculosis in adult'i a clinicopathological study based on sixty-three cases diagnosed at autopsy. Kew Eng J \Ied 23.5: 239, 1946 15 Treip C, \Ieyers D: Fatal hlberculosis in a general hospital a diagnostic problem. Lancet 1: 164, 19.59 16 Rahin CB: Radiology of the chest, ill Golden's Diagnostic Roentgenology, Robhins LL (ed) Baltimore, Williams and Wilkins Co. 1968 p 61 17 jenkins DE, Wolinsky E: \Iycohacterial diseases of the lung and hronchial tree: clinical and lahoratory aspt'cts of hlherculosis. In Texthook of Pulmonary Diseases, Baum GL (ed) Boston, Little, Brown and Co, 196.5 p 152 Rf'print request,,: Dr. Berger, \It. Sinai Hospital Sf'rvices, 7901 Broadway, Elmhurst 73, ;'\ew York.