Detection of Microorganisms in Granulomas That Have Been Formalin-Fixed: Review of the Literature Regarding Use of Molecular Methods

Granuloma is an organized aggregate of immune cells that under the microscope appear as epithelioid macrophages. A granuloma can only be diagnosed when a pathologist observes this type of inflammation under the microscope. If a foreign body or a parasite is not observed inside the granuloma, stains for acid-fast bacilli and fungi are ordered since mycobacteria and fungi are frequently the cause of this type of inflammation. It is calculated that 12 to 36% of granulomas do not have a specific etiology and many have wondered if with new molecular methods we could reduce this number. This paper will summarize the frequently known causes of granulomas and will present the recent literature regarding the use of molecular techniques on tissue specimens and how these have helped in defining causative agents. We will also briefly describe new research regarding formation and function of granulomas and how this impacts our ability to find an etiologic agent.


Introduction
Granuloma is an organized aggregate of immune cells that under the microscope appear as epithelioid macrophages [1]. Epithelioid macrophages are enlarged phagocytic cells that have abundant cytoplasm and can sometimes coalesce to form multinucleated giant cells. e epithelioid macrophages and giant cells are accompanied by other in�ammatory cells including lymphocytes, plasma cells, and polymorphonuclear leukocytes as well as varying degrees of necrosis. e process may be con�ned forming a granuloma or can in�ltrate surrounding tissue thus being called granulomatous in�ammation. is chronic in�ammatory response has been attributed to delayed hyperimmune reaction to a persistent noxious stimulus. �raditionally, granulomatous in�ammation is thought to wall of the noxious agent. However, new experimental observations in the formation of granulomas are now emerging and giving new insides as to how they form and if in reality they are encasing the agent that causes them.
A granuloma or granulomatous in�ammation can only be diagnosed when a pathologist observes this type of in�ammation under the microscope. If a foreign body or a parasite is not observed inside the granuloma, stains for acid-fast bacilli (AFB) and fungi (GMS) are ordered as mycobacteria and fungi are frequently the cause of this type of in�ammation. ere are other diagnostic entities that need to be entertained depending on the tissue, but it is customary to rule out mycobacteria and fungi �rst before considering other diagnostic possibilities. It is calculated that 12 to 36% of granulomas do not have a speci�c etiology [2,3] and many have wondered if with new molecular methods we could reduce this number. is paper will summarize the frequent known causes of granulomas and review the recent literature regarding the use of molecular techniques on tissue specimens that have helped in diagnosing the cause. We will also brie�y describe new research regarding formation and function of granulomas and how this impacts our ability to �nd an etiologic agent.

Causes of Granulomas
From the morphologic perspective, granulomas can present different amounts of necrosis, different sizes and shapes (stellate or geographic), different amounts and types of accompanying in�ammatory cells (suppurative, plasma cells), or the macrophages may show some characteristic morphology (foamy macrophages). Tuberculosis is the most frequent cause of granulomatous in�ammation worldwide. �ther infectious causes include fungi (histoplasmosis, cryptococcosis, blastomycosis, coccidioidomycosis and some of the molds), parasites (schistosomiasis, leishmaniasis, fasciolosis, enterobiasis), and other bacteria (non-tuberculoid mycobacteria, cat scratch diseases-bartonellosis, tularemia, syphilis, listeriosis, Q fever). Viruses are rarely mentioned as causes of granulomas. Some of the morphologic characteristics tend to be associated with speci�c organisms and Table 1 presents a correlation between different granuloma morphologies and the usual etiologic agent encountered [2,4].
e correlation of a speci�c organism with a characteristic granuloma morphology can vary, particularly in immunosuppressed individuals. For example Bartonella henselae can cause cat-scratch disease which is characterized by stellate granulomas but in immunosuppressed patients this bacteria can cause bacillary angiomatosis-peliosis [5]. In HIV infected patients, cutaneous histoplasmosis can have mononuclear and plasma cell in�ammatory in�ltrate in the super�cial and deep dermis with perivascular, perifollicular, and lichenoid patterns and not a granulomatous in�ammatory pattern [6]. Although granulomatous in�ammation is the usual morphology described for patients with Mycobacterium tuberculosis, there have been some descriptions of cases in which the in�ammatory response is not granulomatous. A study of necrotizing nongranulomatous lymphadenitis showed presence of M. tuberculosis in 6 out of 35 cases by using molecular methods in formalin-�xed, paraffin-embedded tissues [7].
Granulomas or granulomatous in�ammation due to an infectious agent can be associated with other pathologies; thus it may be difficult to recognize it as independently caused entities. e collision of two entities has been documented in HIV patients that have in the same lesion Kaposi sarcoma and granulomas with presence of acid-fast bacilli and molecular evidence of mycobacteria [8]. Patients with breast cancer and tuberculosis in the same lymph node have been described [9].
In some instances there is a granuloma and an infectious agent is found; however, this is not a typical organism that causes granulomas such as Pneumocystis [4,10]. In these instances the pathologist wonders if other causes should be searched for; however, review of the literature of cases with Pneumocystis and granulomas has not shown presence of other organisms. e variable histopathologic pattern in the presence of the same organism suggests that host factors are at play [11,12]. e host factors that have been proposed include presence of CD4 and CD8 lymphocytes, previous Pneumocystis exposure, absence of IgA antibodies against Pneumocystis, and the exposure time to the fungal glycoproteins. e role played by the immune system in formation of granulomas is further emphasized when considering some congenital immunode�ciencies. e typical example is chronic granulomatous disease which is a childhood disease involving defects in the phagocytic function as there is reduced nicotinamide adenine dinucleotide phosphate (NADPH) oxidase complex. ese patients mount granulomatous in�ammation towards most infectious agents and even noninfectious etiologies [13]. Other immune de�ciencies such as common variable immune de�ciencies can also present granulomatous in�ammation towards infectious agents that are not usually associated with granulomas [14].
ere are noninfectious causes of granulomas. Foreign bodies such as talc, beryllium, mineral oils, lipids, or medications can produce granulomatous in�ammation in diverse organs. Sarcoid-type granulomas are also seen in lymph nodes of patients with lung, stomach or uterine neoplasias [15][16][17] or may appear in the skin preceding or concomitant with the initial diagnosis of cancer [18]. e presence of granulomatous in�ammation may confer better prognosis in some neoplastic diseases as it may indicate some degree of host antitumor activity. is has been documented in patients with small cell carcinoma of the lung in whom there is sarcoid-like granulomatous in�ammation of regional draining lymph nodes [19]. Hodgkin's disease is the neoplasia most frequently cited to cause diagnostic confusion because of multisystem granulomatous in�ammation which can have various degrees of necrosis [15]. A variety of immune dysfunctions associated with the neoplasia such as a systemic in�ammatory reaction with production of chemokines, interleukins and angiogenic factors, immune cellular changes such as anergy, and epigenetic changes such as hyper-methylation have been implicated in the formation of the granulomatous in�ammation [18]. Nowadays with the use of immunomodulators such as interferon-and monoclonal antibodies for different neoplasias and other infections, granulomatous in�ammation has been seen a�er these treatments with no evidence of an infectious agent [20,21].
ere are entities that cause granulomas and are speci�c for certain organs. For example, in the upper respiratory tract Wegener's ganulomatosis, Churg-Strauss syndrome, or cocaine-induced midline destructive lesions [22] need to be considered. In the liver the entities that need to be considered include primary biliary cirrhosis, primary sclerosing cholangitis, hepatitis C, and others [2,3]. Also to consider is the variable distribution of granulomas in different tissues in the various entities. For instance, although we think of sarcoid as having most of the lesions in skin, respiratory tract, and lymph nodes [23], 80% of patients have been shown to have microscopic hepatic granulomas even though symptomatic liver disease is not frequent [24]. From the pathologic perspective we think of Brucella-associated hepatic �brin ring granulomas� however, these are only seen in 2 to 10% of cases [2] while vertebral and osteoarticular involvement is probably much more common and the usual causes of mortality include endocarditis and central nervous system involvement [25,26].
e frequency of the different etiologies for the granulomatous in�ammation will depend on the prevalence of the diseases in that particular geographical area. For example, a study in Southern Iran of causes of liver granulomas showed that a little over 50% of their cases were due to tuberculosis and 8% were due to visceral leishmaniasis while other etiologies included fungal infections, visceral larva migrans, primary biliary cirrhosis, hepatitis C, cytomegalovirus, lymphoma, and others [3]. In this study they used a variety of methods to de�ne the causative agent including histopathology with special stains, PCR, autoantibodies, viral markers, and other characteristics obtained from the chart review which were not always evident at the time the hepatic biopsy was studied. A study in Turkey of hepatic granulomas found that 44% of their patients had different stages of primary biliary cirrhosis, while only 12% had tuberculosis (all 10 patients diagnosed with PCR and one having AFB positive staining), 6% had neoplasias (primarily cancers of hepatic origin), and 4% had sarcoidosis [27]. An Egyptian study of cutaneous granulomas showed that tuberculosis accounted for 40% of cases followed by leprosy (32%), leishmaniasis (16%), while actinomycetes and fungi were rare [28]. Diagnostic methods for the Egyptian study included intradermal tests, biopsy, serology, direct smear from lesion and PCR.

Mycobacterial Detection Using Molecular Methods in Patients with �ranulo�atous �n�a��ation
Mycobacterial DNA can be recuperated and the species iden-ti�ed using molecular methods in a variety of samples such as sputum, cultures, formalin-�xed, para�n-embedded tissue blocks that have been used for histopathology, and bones of mummies from Egyptian times [29]. Different primers have been used including: (a) the 123-base pair (bp) segment of the repetitive IS6110 sequence of M. tuberculosis complex, which covers M. tuberculosis, M. bovis, M. microti, M. africanum, and M. canettii [30][31][32], (b) EUB338 [33], (c) MTB187 [33], (d) MAVP187 [33], (e) 16S rDNA [34], (f) hsp65 [35], (g) KD1 and KD2 [36], (h) 16SOL-16SOR and 16SIL-16SIR [37], and (i) pncA (Rv2043c) [38]. For the most part, DNA from different specimens is extracted then PCR reactions are performed to detect mycobacteria. Con�rmation that the product is M. tuberculosis or another mycobacteria has been done by a variety of methods including sequencing and Southern blot. Other methods that have been used to test cultures or formalin-�xed, para�n-embedded tissue sections are peptide nucleic acid �uorescent in situ hybridization [39], rRNA oligonucleotide probes [33], and spoligotyping (a hybridization step a�er IS6110 PCR ampli�cation that allows differentiation of the nonrepetitive DNA spacers of M. tuberculosis and bovis) [32]. Immunohistochemistry has been used to detect mycobacterial antigens in tissue sections and some authors have found that this technique has better sensitivity than molecular methods in addition to being easier to perform in resource poor settings [40]. However, with immunohistochemistry there is cross-reactivity between the different mycobacteria species [41].
It is important to realize that in many occasions a tissue sample is not sent for cultures since a granuloma or granulomatous in�ammation is not found until microscopy is performed and all the tissue, by then, has placed in formalin. Also, a sample may have been sent for culture but neither mycobacteria nor fungi grew from the specimen sent to microbiology, thus the only material available for testing is the formalin-�xed, paraffin-embedded tissue. Following we will describe the studies that have focused on studying formalin-�xed, paraffin-embedded material exclusively.
(1) A study of 48 formalin-�xed, paraffin-embedded tissues with granulomas in lung, bone, lymph nodes, liver, genitourinary tract, and other tissues showed positive AFB staining in 8 (17%) cases and molecular evidence of mycobacteria in 19 (40%) [32]. e authors used the repetitive IS6110 sequence of M. tuberculosis complex to de�ne the presence of mycobacteria and spoligotyping was used to differentiate between the different species. e authors identi�ed M. tuberculosis in 7 cases, M. bovis in 4 and in 9 cases the spoligotypic was not able to de�ne the mycobacteria present. Of the cases that were AFB positive, M. tuberculosis was found in 5 cases, M. bovis in 1, and in 2 the determination of the mycobacterial species was not possible.
(2) Another study of 190 cases with granulomas from the respiratory tract, lymph nodes, and other tissues targeted the hsp65 gene in a nested PCR to detect all mycobateria then sequenced or used restriction fraction polymorphisms for species identi�cation [35]. In some cases culture results were available and concordance of molecular and culture results was found in 83% (34/41 cases). A total of 119 (63%) cases were positive for mycobacterial DNA. e typical restriction pattern for M. tuberculosis complex was present in 71 cases (60%) while non-tuberculous mycobacteria were found in 41 cases (34%) and identi�cation was not possible in 15 cases (36%). Only 7 cases showed AFB positive staining out of 14 that had M. tuberculosis by culture and molecular methods.
(3) In a study from ailand of 120 skin cases with granulomas in which cultures had been obtained, PCR using 16S rRNA common fragment for mycobacteria and a second 306 bp DNA fragment speci�c for M. tuberculosis was performed [37]. AFB staining in formalin-�xed, paraffin-embedded sections showed positivity in 31%, by PCR 36% were positive and by culture 30%. ere were 12 cases that were positive with both AFB stains and PCR, 21 that were AFB positive but PCR negative, and 35 that were PCR positive but AFB negative. e sensitivity and speci�city of PCR against AFB results was 29 and 61%, respectively, and when compared to culture these were 67 and 76%. e low number of cases with positive results for PCR was attributed to inhibitors present in the tissues.
(4) A study from Hong Kong of 115 patients correlated histopathology, culture and PCR results [30]. e authors used the IS6110 target and con�rmation of M. tuberculosis was performed by Southern blot hybridization. Between 1 to 3 paraffin blocks per patient were tested as they assumed that mycobacteria were scant and randomly distributed in the tissues. ey looked at granulomatous in�ammation of 62 pulmonary specimens and 53 extrapulmonary sites. Only 68 patients had PCR and culture results. e authors showed that 20 patients were positive using culture and PCR, 2 were PCR negative but culture positive, and 13 were negative for culture and PCR. An additional 35 cases were positive with PCR but no cultures were available for them. Positive AFB staining and PCR occurred in 12 cases.
(5) A study from Korea looked at 250 patients with culture and AFB smears that required lung biopsies because the clinical diagnosis of tuberculosis was uncertain [31]. In this study the authors calculated sensitivity and speci�city of nested PCR using the IS6110 target against culture to be 85 and 99%, respectively.
(6) A study of 50 Indian patients with abdominal tuberculosis showed 31 cases positive with histopathology and 30 with a PCR that used the KD1 and KD2 target [36]. Only 24 cases were concomitantly positive with histopathology and PCR and 11 that were negative with both techniques. In 4 patients the PCR did not yield results due to inhibitors.
(7) A study of 12 Japanese patients with granulomatous in�ammation of musculoskeletal tissues used the 16S rDNA and sequenced the products of the formalin-�xed, paraffin-embedded tissues [34]. ey obtained amplicons from 5 cases (41%) and the mycobacteria found included M. tuberculosis in 4 cases and M. avium in one. Four of these cases showed AFB in the tissues, 3 cases had M. tuberculosis and one M. avium.
(8) A second group from Korea correlated histopathology, AFB staining and a nested PCR using the repetitive IS6110 sequence of M. tuberculosis complex in 81 patients [42]. ey studied a variety of tissues from the respiratory tract, lymph nodes, gastrointestinal tract, so tissues, and others. ey had 53 cases with granulomatous in�ammation in which 17 were AFB positive and 36 PCR positive. Of the 17 AFB positive cases 14 were also PCR positive thus 3 cases were attributed to other mycobacteria. Of the 28 cases with chronic in�ammation but no de�nitive evidence of granulomas none showed positive AFB while 10 showed evidence of M. tuberculosis by PCR.
(9) A study from Spain of 10 patients with papulonecrotic tuberculid skin lesions demonstrated that 8 had M. tuberculosis complex in formalin-�xed, paraffinembedded skin biopsies [43]. e authors used a 285bp sequence speci�c of M. tuberculosis complex that was con�rmed by Southern blot hybridization. In this study the patients did not have respiratory evidence of tuberculosis but all were PPD positive.
Although these studies have used an array of different clinical presentations, tissues, and pathologies, they have Scienti�ca 5 established that PCR can detect mycobacteria in 36 to 60% of formalin-�xed, para�n-embedded specimens. In many instances staining AFB in the tissue was not observed. If the primer used in the study was speci�c for M. tuberculosis, presence of AFB staining did not correlate with PCR positivity and in those cases the authors commented that the AFB could correspond to other mycobacteria. From the previous studies that used formalin-�xed, para�n-embedded tissues and others that have used a variety of materials such as fresh samples and cultures it is clear that obtaining a product for PCR analysis using a broad mycobacterial target then sequencing and comparing with known GenBank sequences can reveal that those mycobacteria are not M. tuberculosis but other mycobacteria. Examples of detection of these mycobacteria are presented in Table 2.
Detection of mycobacterial species does not seem to improve if fresh tissues rather than formalin-�xed, para�nembedded tissues are used. A study from India that used fresh tissue to perform PCR for M. tuberculosis targeting the IS6110 gene and compared to histopathology (necrotizing granulomas) showed that 20 of 104 (19%) samples were both positive with AFB and PCR while 74 (71%) were negative for both [44]. Discrepant results occurred in 7 cases, 4 of these were positive with the AFB stain but negative with PCR and 3 were positive with PCR but negative with AFB.
Molecular techniques are expected to increase sensitivity and speci�city� however, in some instances this does not occur. A study of bovine lymph nodes with Mycobacterium bovis used the IS1081 which detects down to 1 genome copy and the RD4 which detects 5 genome copies. e authors determined a sensitivity of 70% and 50%, respectively, for each probe compared to culture [45]. When analyzing why the recovery with PCR was lower than expected, the authors determined that there was limited recovery of mycobacterial DNA which may have been due to the resilient mycobacterial cell wall, the presence of tissue debris and the paucibacillary nature of some of the lesions. In order to decrease the amount of tissue debris researchers have compared PCR on laser capture microdissection material to the entire formalin-�xed, para�n-embedded tissue sections but found no improvement in sensitivity and speci�city for detection of M. tuberculosis [46].
In some instances the organism detected by the molecular technique may not be the same obtained in cultures. A study of resected lung specimens from 24 patients with nontuberculous mycobacterial disease correlated the organisms identi�ed by culture of tissue or sputum to the results obtained from molecular testing using IS6110 to detect M. tuberculosis and hsp65 gene which was later sequenced [47]. In 12 patients the culture and sequencing analysis rendered the same organism. In 8 patients the culture and PCR results were discrepant. In 6 of the patients the culture showed a rapid grower. Ten of the patients had �brocavitary disease and in all but 2 more than one mycobacteria were identi�ed. e authors suggested that the frequent discrepancies between the PCR and culture results were possibly due to multiple non-tubercoulus bacteria colonizing the lesions. e authors comment that when using PCR, the most abundant mycobacteria is the one identi�ed although in some instances more than one mycobacteria were observed. In contrast, the organism that overtakes the culture is the one growing faster.
By use of molecular techniques new species of mycobacteria are being discovered. For example, a case of skin papules and nodules that showed granulomatous in�ammation in a patient with Hodgkin's disease and severe cellular immun-ode�ciency demonstrated Mycobacterium simiae when using the sequence of the product of 16S rRNA. However, based on culture characteristics, susceptibility testing, molecular testing targeting different portions of the mycobacteria gene (rpoB and hsp65), and high-performance liquid chromatography of mycolic acid methyl esters the mycobacteria was determined to be a novel one and the authors proposed this to be called M. shigaense [48].
Molecular testing has also been useful in study of nosocomial infections as it can de�ne the species of mycobacteria and other mutations that can occur. For example, random ampli�ed polymorphic DNA-PCR was used to de�ne clonicity of Mycobacterium abscessus in isolates from 8 patients that had granulomatous prostatitis aer biopsies had been performed on them [49]. All isolates were indistinguishable from one and other suggesting that all isolates came from the same clone and had been nosocomially acquired.

Detection of Organisms Other than Mycobacteria Using Molecular Methods in �atients �ith �ranulomatous �n�ammation
Our currently used culture and staining techniques are optimized to identify microbial agents that are known to cause disease. Similarly, detection of antigens, antibodies, or presence of nucleic acids are targeted to identify a particular suspect disease. Table 3 presents methods that can be used for diagnosis of infectious agents not related to mycobacteria that cause granulomatous in�ammation.
To discover new pathogens, molecular methods that use conserved portions of microbial genomes to obtain products which are then sequenced are ideally suited for this task. For bacteria and fungi, some portions of the ribosomal nucleic acids are highly conserved and can be used to determine the genus and species. Tropheryma whippelii is a clear example of discovering an agent for a disease that had long been thought to be infectious but in which cultures had failed to give us the microbial agent [97]. e discovery was based on obtaining a product using a 16S ribosomal RNA (rRNA) primer and then sequencing the product. Sequencing the 16S rRNA product of bacillary angiomatosis lesions demonstrated that B. henselae (previously Bartonella bacilliformis) was the causative agent [98]. In the case of T. whippelii the genetic sequence was not recognized thus this was a new distinct bacterium while comparing the sequence of B. henselae to others published at the time demonstrated that the bacteria present in bacillary angiomatosis lesions was related to Bartonella and Rochalimaea. As described previously, this conserved nucleic acid sequence of bacterial small rRNA has been used for detection of different mycobacteria species. In the case of fungi, products of conserved rRNA genes (18S, 28S and 5.8S) or the intervening internal transcribed spacer (1 → 3) -D-glucan has a diagnostic sensitivity of 95% but speci�city is 86% since there is cross-reactivity with other fungal infections.
Not available.
Urine and serum antigen cross-react with Blastomyces. Antibodies can also be measured.
Can take up to 4 weeks to grow. Can be found in blood cultures.
Urine and serum antigen cross-react with Histoplasma. Antibodies can also be measured.
Grows well but may take several weeks. Growth may take up to 2 months.
(ITS1 and ITS2) regions can be obtained, sequenced and the latter compared to what has been published to help determine the fungal genus and species present in the tissue [84,99,100]. A retrospective study of fungi in formalin-�xed, paraffinembedded tissues using conserved primers and sequencing the product obtained showed that nucleic acid recovery varied from 60 to 90% depending on the method used for extraction and led to a PCR efficiency of 57 and 93% which translated to approximately 60% recovery of the fungi to the genus level [101]. For parasites and viruses, widely conserved regions are not available.
Even with these advances, we are constantly confronted with cases in which an infectious agent is suspected but no speci�c agent can be found. A typical entity falling in this category is sarcoidosis which has always been suspected to be of infectious origin but in which �nding the causative organism has been elusive [23]. ese patients have granulomas with little necrosis and no mycobacteria or fungi are observed in the tissue by using conventional techniques or evidence of a microorganism is found by a variety of methods (culture, skin testing) at the time of original diagnosis. New molecular methods and immunologic assays have been used to study a variety of sarcoidosis samples and the association with a variety of mycobacteria appears to be strong [23,102]. Using Mycobacterium species 16S rRNA and rpoB sequences, researchers have found that 60% of lung biopsies and lymph nodes pathologically classi�ed as sarcoid contained mycobacteria; however, the IS6110 speci�c for M. tuberculosis was negative [103]. is study found a heterogeneous group of mycobateria that are closely related to M. tuberculosis, M. gordonae, and M. kansasii. e amount of mycobacteria using real-time quantitative PCR targeting the IS986 of the M. tuberculosis genome in formalin-�xed, paraffin-embedded specimens of patients with sarcoidosis is present in low numbers compared to controls [104]. Moreover, IgG antibodies against M. tuberculosis catalaseperoxidase have been found in a high proportion of patients with sarcoidosis compared to controls [105].
In addition to mycobateria, other organisms including Propionibacterium acnes have been found in greater amount in specimens from patients with sarcoidosis by molecular analysis too [23,106]. e literature that links mycobacteria with sarcoidosis has primarily been observed in American subjects while that linking Propionibacterium with sarcoidosis is in Japanese persons, suggesting that host genetics play an important factor regarding the possible causative agent to what we now call sarcoidosis. Even when organisms, in particular mycobacteria, are isolated or found in patients in whom a previous diagnosis of sarcoidosis had been given, the question of whether the mycobacteria was the cause of the granulomas or if there are concomitant diseases (sarcoidosis and an infection with a mycobacteria) is always present [102].
Nowadays a revolution in microbiology laboratories is taking place with the use of matrix-assisted laser des-orption�ionization time-of-�ight (MALDI-TOF) mass spectrometry analysis [107,108]. is method studies the composition of biomolecules in bacteria including proteins, DNA, sugars and others. e �ngerprint of a growing number of bacteria are now in the libraries of MALDI-TOF mass spectrometers. For this method, bacterial colonies are placed in a matrix and allowed to cocrystallize with the matrix as it dries. en the sample is �red with a laser causing ionization of the biomolecules. e ionized particles are pulsed into a time-of-�ight mass spectrometer. Timeof-�ight mass spectrometers have a �mirror� that re�ects the ions and allows for better discrimination between the biomolecules measured creating a �ngerprint pattern speci�c for the bacteria. en the pattern of the suspect organism is compared to a library of patterns and identi�cation to the genus and species level can be achieved based on the speci�c �ngerprint of biomolecules. For some organisms the discrimination of the �ngerprint has a success that ranges between 89 to 100% for the genus and between 84 to 95% for the species. Since identi�cation is based on the organisms available in the library, as a wider diversity of organisms are entered into the libraries of the different instruments in the market identi�cation rate will improve. At this point, the instrumentation is expensive; however, the identi�cation of each sample is cheap. �acterial identi�cation is much faster once a pure colony has been isolated since bacterial growth with production of a variety of biochemical reactions is not needed which will be a major improvement in turnaround time for identi�cation of mycobacteria. It should be noted that for some organisms an extraction phase is necessary. Study of primary specimens using MALDI-TOF is being pursued for sterile sites such as blood, cerebrospinal �uid, and �ne needle aspirates which may allow discovery of microbiologic agents that are fastidious growers and cause granulomatous in�ammation.
Lastly, there may be some granulomatous in�ammatory processes in which a causative agent may never be found and some authors have suggested that host factors play a mayor role. A Canadian study of central nervous system granulomas with caseous necrosis but in which an etiologic agent could not be determined, demonstrated no improvement or worsening of clinical and radiologic features aer treating with antimycobacterial regiments but improvement aer immunomodulation therapy with prednisone or azathioprine [109]. In this study, search for causative agents, primarily mycobacteria, included a variety of stains, cultures, and PCR and the authors suggested that if no organism is found the most probable cause is not infectious but a variant of a hypersensitivity reaction, possibly sarcoidosis, where granulomas show necrosis.

Formation and Function of Granulomas
Traditionally granuloma formation has been thought as an adaptive cellular immune process. ere are several phases in the formation of granulomas: initiation, accumulation, effector, and resolution [24]. Initiation takes place when an antigen is presented by a macrophage or dendritic cell to CD4 lymphocytes. e presentation is usually done through use of Toll-like receptors that recognize a variety of structural components such as mycolic acids and peptidoglycans present in the cell wall of mycobacteria and other granuloma inducing agents [110]. T helper 1 (1) CD4 cells are usually involved in mycobacterial diseases and sarcoidosis, while T helper 2 (2) CD4 cells are usually involved when there is a parasitic infection. As infection progresses, there can be switching from 1 to 2 cell response. 1 cells produce Il-2 and interferon (IFN ) which recruit more CD4 cells and macrophages and are indispensible for the accumulation phase [110]. Other chemokines that play a role include tumor necrosis factor (TNF), and several interleukins produced by 2 cells (IL-4, IL-5, and IL-13) and macrophages (IL-12). e 1 response induced delayed hypersensitivity reactions while the 2 response stimulates antibody production, in particular IgE [111]. e new milieu recruits CD8 lymphocytes and neutrophils and is important for the effector phase. Lastly the resolution phase includes the deposition of collagen and creation of �brosis. is conventional thought regarding the formation of granulomas results in walling off the noxious agent.
Recent experiments have challenged some of the concepts regarding the production of granulomas. Following we will explore how our understanding of different components has evolved with respect to granulomas formed by different agents. Neutrophils in granulomatous in�ammation have an active role [112]. Experiments using intradermal injections of Bacille Calmette-Guerin (BCG) have shown that neutrophils, rather than macrophages, are the cells that carry the mycobacteria into the lymph nodes. Once in the lymph nodes, neutrophils present mycobacterial antigens to dendritic cells that then recruit CD4 cells and lead to production of IFN . In addition, neutrophilic proteins are present in macrophages located in ganulomatous in�ammation. It is postulated that these proteins serve to enhance phagocytic properties of macrophages. Although the important role of neutrophils is histopathologically con�rmed in Bartonella, Listeria, Candida, and Toxoplasma granulomatous in�ammation [58], neutrophils have not been traditionally considered an important component in mycobacterial infections.
Other researchers have demonstrated that innate immunity is an important component of granulomatous in�ammation. Using zebra�sh larvae which do not have adaptive immunity, researchers have shown that granulomas form when they are exposed to Mycobacterium marinum [1]. e zebra�sh experiments have also demonstrated that macrophages that have ingested the infectious agent during early infection are responsible of propagating the infection to other macrophages as they undergo apoptosis rather than containing the infectious agent [113]. Macrophages forming granulomas move rapidly towards infected macrophages and ingest mycobacteria. is exchange usually occurs at the periphery of the granuloma. e signal for macrophage movement is derived from the pathogen and the dying infected macrophage. As the newly recruited macrophage ingests the contents of the dying one, it allows for mycobacteria to multiply and disperse with the moving macrophage. In this manner, the infection is not walled off but disseminated from a primary site to a secondary one. In addition to macrophages moving fast through the granuloma, T cells are also moving. us, it is postulated that there are 2 types of functional granulomas: those with rapid moving macrophages which are formed in response to infectious agents such as mycobacteria and those with slow moving macrophages which include those responding to foreign bodies [1].
Review of histopathologic descriptions of tuberculosis in the preantibiotic era comment that primary infection resulted in an exudative reaction which included foamy alveolar macrophages and some degree of lipid necrosis and systemic dissemination of the mycobacteria [114]. Using molecular techniques, some investigators have corroborated that mycobateria are disseminated through the body by using PCR and spoligotyping [115]. e mycobacteria have been found during latent infection in vertebrae [29,116], spleen, kidney and liver [115]. By using in situ hybridization, the mycobacteria have not only been present in macrophages, but also in endothelial cells of these organs [115]. Of interest is that in the same patient more than one bacterial genotype was found.
Our understanding of the role that different mediators have has also evolved through time. For example, it appears that TNF is not indispensable for the formation of granulomas due to mycobacteria in zebra�sh. However, the role of matrix metalloproteinase 9 (MMP9) secreted by epithelial cells is very important for recruiting macrophages and forming granulomas [117]. MMP9-knockout mice that have been infected with M. tuberculosis have poor granuloma formation and low amounts of mycobacteria.
Besides the host factors discussed, there are new discoveries that involve mycobacterial factors. For example, it is now recognized that there are several cell wall lipids that arrest the maturation of phagosomes and do not allow acidi�cation [118]. e shi from an external pH of 7 to the phagosomal pH of 6.4 shapes the transcriptional response of mycobacteria allowing them to continue to grow inside the macrophage by using host cholesterol and other cell lipids as nutritional sources. In addition, mycobacteria avoid destruction by superoxide by using superoxide dismutase and by cell wall lipidoglycans scavenging the oxygen radicals. e mycobacteria residing inside the phagosomes release proteins and lipids that traffic actively through the infected cell. As a pathologist one would expect abundance of bacilli in these lesions but the organisms are usually difficult to �nd. One explanation to pathologist�s inability to �nd mycobacteria may be due to an incompletely formed cell wall during the replicative process which does not allow visualization with the techniques we currently use. Some researchers have studied the lipid composition of the caseous necrosis in tuberculoid granulomas and have found that there is dysregulation of lipid metabolism in macrophages with disproportionate abundance of certain lipid metabolic pathways [119] which explains why there is a lot more caseum than bacteria in them. Another possible explanation is that mycobacteria become cell wall de�cient due to the host immune response and the shape and acid-fast staining of the mycobacteria is atypical or lacking [120,121].
Microscopically, acid-fast bacilli are usually found in the interface between necrotic cellular debris and areas with caseous necrosis. As the granuloma becomes older, bacteria become sparser in the areas of hard necrosis. Review of histopathologic descriptions of tuberculosis in the preantibiotic era comment that post primary tuberculosis lesions start with foamy macrophages that have ingested the organisms [114]. In 90% of the cases the lesions regress spontaneously and �brosis is noted in the area where the foamy macrophages were, but in 10% of the cases the lesions evolve to caseous necrosis and cavity formation. e numbers of acid-fast bacilli in these cavities are increased and as the individual coughs, other persons are at risk of acquiring the disease.
A similar distribution of mycobacteria has been found in granulomas of immune competent patients caused by Mycobacterium avium complex [122]. During initial infection, when only one exudative lesion is seen, there are many organisms inside mononuclear cells. As the lesions progress to a well de�ned granuloma with necrosis, there are less organisms which are usually present in the periphery either inside mononuclear cells or in the necrotic material.
Schistosoma produced granulomas in humans show a mixed 1-2 response that will vary depending on the genetic background of the host, nutritional status, intensity and duration of the infection, and concurrent pathology [111]. In murine animal models the 2 response with a large population of eosinophils dominates the response to schistosome eggs. e 2 response with production of IL-10 and IL-13 favors �brosis but, this response may also depend on other compounds such as nitric oxide synthase. e 1 response with production of TNF, IFN and IL-12 is responsible for anti�brotic activity [111,123]. Also in play is an increase in Treg cells that modulate the 1-2 response. In immunocompetent hosts, Schistosoma eggs elicit a collagen-rich granulomatous response to try to sequester egg products, but this same response can lead to extensive hepatic �brosis. �nderstanding the pathobiology of this parasitic infection gives us insights into the difficulties that can occur when in a liver biopsy, granulomas with eosinophils and extensive �brosis are found but no egg parasite can be identi�ed. To further de�ne what is causing the disease in each individual patient requires epidemiologic information that will guide further serologic, molecular and other studies. Studies of formation of granulomas in vitro using peripheral blood cells from patients with acute and chronic Q fever and healthy controls have corroborated some of the �ndings from the zebra�sh and shed some insights into differences encountered at various stages of the disease [124]. In their experiments, monocytes from healthy donors were the ones that initiated granuloma formation but well formed granulomas required the presence of T lymphocytes. In addition to monocytes and lymphocytes, their granulomas had epithelioid macrophages, and multinucleated giant cells, but no dendritic cells. Distance and trajectories covered by monocytes and T lymphocytes of healthy donors to Coxiella coated beads seems to be similar to that found in zebra�sh� however, these were diminished if the monocytes and T lymphocytes came from patients with chronic Q fever. In addition, monocytes and T lymphocytes from chronic Q fever were either unable or displayed partial granuloma formation. e authors comment that their experiments recapitulate what happens in vivo since patients with acute Q fever form �brin ring granulomas but as the infection becomes chronic the granulomas are defective and only spotty necrosis surrounded by collections of lymphocytes are observed.

Conclusions
Granulomatous in�ammation is a complex immune reaction that includes both innate and adaptive immunity and can be caused by a variety of infectious agents. When the granuloma is caused by mycobacteria, disseminated disease has now been well documented using molecular methods. Molecular methods have also been very useful to de�ne that not all granulomas with AFB are due to M. tuberculosis and that mycobacteria are sometimes found in the absence of bacilli. However, an etiology for granulomatous in�ammation with or without necrosis is not always found, even with the use of molecular methods. De�ning which granulomas need to be tested using molecular methods, which probes or primers and protocols to use routinely, and what is the ideal specimen is not clear at this point as the series presented vary largely in these aspects.