Our data suggest that treatment failure or recurrence of pulmonary actinomycosis is not uncommon. The 1-month antibiotic response was an independent predictor of the treatment outcome. This is the largest published series of biopsy-proven cases of pulmonary actinomycosis and is to our knowledge the first study to evaluate the treatment outcome using a multivariate analysis. The mean follow-up period of 37.7 months was also longer than in previous pulmonary actinomycosis studies.
In our series, the classic presentation of pulmonary actinomycosis with chest wall invasion and cutaneous fistulas discharging sulfur granules was absent. This was common in the pre-antibiotics era, but is now rare . The change might be the result of improvements in oral hygiene, the ready availability of imaging modalities such as CT, and the early initiation of antibiotics when pulmonary infection is suspected . A male predominance was noted and there were a large number of smokers and drinkers. These are consistent with the findings of earlier studies [6, 11]. COPD and previous mycobacterial infection were the most common underlying respiratory disorders, and some patients had diabetes mellitus, as in previous studies [6, 12]. However, it is not clear whether an immunocompromised status is a risk factor for actinomycosis. For example, the prevalence of actinomycosis was not increased in patients with acquired immunodeficiency syndrome . In fact, the majority of patients in our study were immunocompetent. Education status also did not influence the incidence. A previous study also documented that socioeconomic class per se did not appear to correlate with the incidence of actinomycosis in the developed world .
The definitive diagnosis of actinomycosis needs the direct isolation of Actinomyces from a clinical specimen. However, cultures are positive in only a few cases because of failure of the anaerobic organism to grow . The diagnosis is often based on histopathology showing Gram-positive filamentous branching bacteria and sulfur granules. All of our patients were diagnosed from biopsies. Microbiological cultures were performed for the biopsy specimens in 26 patients. However, the results were positive in only one patient.
Our treatment outcome for pulmonary actinomycosis consisted of a cure rate of 86.8% (59/68), failure rate of 8.8% (6/68), and relapse rate of 4.4% (3/68). There was no mortality from pulmonary actinomycosis. One patient died of metastatic gastric cancer 3 years after finishing successful treatment. Previously, our group reported 25 cases of pulmonary actinomycosis, and one patient died of a bronchopleural fistula (BPF) and empyema after a pneumonetomy . Song et al.  reported an overall cure rate of 85% (34/40) in pulmonary actinomycosis without mentioning recurrence. They also experienced one case of postoperative mortality due to a BPF. Choi et al.  reported a 100% cure rate in 26 cases without recurrence, despite a relatively short follow-up duration (median 23 months). Possible explanations for treatment failure are drug-resistant co-pathogens and poor penetration of the drug caused by avascularity and induration of the infected area . Although reported antimicrobial susceptibility tests indicate that most Actinomyces are susceptible to penicillin and amoxicillin, Actinomyces may acquire antibiotic resistance during antibiotic treatment [15, 16].
There are no guidelines regarding the appropriate duration of antibiotic treatment. Recently, several investigators reported that shorter courses of treatment could be successful in pulmonary actinomycosis . However, Kolditz et al.  reported that antibiotic treatment for less than 3 months in medically treated patients might be associated with local complications or recurrence. We managed cases of pulmonary actinomycosis with antibiotic treatment for less than 6 months. Furthermore, prolonged treatment did not prevent recurrence in three patients (range 9.2–14.0 months). It is not clear why these cases recurred: one was a current smoker and heavy drinker, but the other two had never smoked and did not consume alcohol. Considering the time to recurrence in our study, a reasonable follow-up duration should be at least 3 years.
Whether IV antibiotics should be used in the early management of actinomycosis remains uncertain. More than half of our medically treated patients were treated successfully without IV antibiotics. Furthermore, the median duration of IV antibiotics was less than 1 week in the group given IV antibiotics. Choi et al.  also reported that 7 of 15 medically treated patients with thoracic actinomycosis were treated successfully with oral antibiotics only. Therefore, the previous recommendation of IV antibiotics for 2–6 weeks might not be appropriate for all patients.
Penicillin G has been recommended as the drug of choice for treatment . However, no comparative study has investigated which antibiotics are superior. To our knowledge, only two in vitro studies have reported Actinomyces drug susceptibility [15, 16]. Recent reviews recommend that the first-line regimen consist of a beta-lactam and beta-lactamase inhibitor [1, 16, 17]. A beta-lactamase inhibitor offers the advantage of coverage against penicillin-resistant aerobic and anaerobic co-pathogens. These microorganisms produce the enzyme beta-lactamase, which can shield Actinomyces from the effect of penicillin. Furthermore, they might assist the spread of infection by inhibiting host defenses and reducing local oxygen tension . In comparisons of AMX-CLV with AMP-SUL, the outcome of treatment with the latter tends to be poor. A possible explanation for this result is that sulbactam is a less potent beta-lactamase inhibitor than clavulanate in vitro , and ampicillin peak levels are delayed and lowered if it is ingested with food . A prospective clinical trial would be the proper approach.
There are no definitive guidelines for when surgical intervention is required. Surgery is needed if a malignancy cannot be excluded . It might also be used to control severe symptoms, such as uncontrolled hemoptysis . Surgery could be a rescue treatment for patients who do not respond to antibiotics . In our study, one of nine patients who presented with massive hemoptysis underwent emergency surgery. Bronchial artery embolization was attempted in the other eight patients, of whom seven ultimately needed surgery for continuing bleeding or for definitive diagnosis. Other indications for surgery include drainage of an abscess or pleural empyema, decortication, and the radical excision of sinus tracts .
One of our most important findings was that the 1-month antibiotic response and treatment outcome were closely associated according to the multivariate analysis. According to previous case reports, pulmonary actinomycosis usually decreases in size on the chest radiograph within 4 weeks . A recent study also reported that only 20% (1/5) of patients with treatment failure showed any response to antibiotics within 4 weeks . In addition, long-term antibiotics could not guarantee a successful outcome or no relapse in our series. The absence of an early radiological response might be an important predictor of a poor outcome. We did not find a significant difference in treatment outcomes based on treatment modality.
This study had several limitations. First, given it's retrospective nature, selection bias may have influenced the significance of our results. The sample size of the study was too small to obtain a strong conclusion. Additionally, many treatment options were used, and this might have influenced the statistical power. Second, there were no predetermined criteria for halting treatment. Antibiotics were stopped in most patients based on radiological and clinical resolution of their pulmonary actinomycosis. Third, we did not evaluate the correlation between drug susceptibility test results and treatment outcome.