The Official ATS/ERS/ESCMID/IDSA Clinical Practice Guidelines recommend treatment with a 3-drug regimen, including a macrolide for macrolide-sensitive MAC-LD . In this study, we firstly reported the problematic adverse events of treatments for MAC-LD by using a large database in Japan.
Of the 1200 adverse event reports, the total number of deaths, non-recovery, and sequelae was 19.4% (233/1200), which cannot be ignored. This result may be influenced by the bias that minor adverse events that are not clinically recognized are unlikely to be reported; therefore, severe adverse events may have been reported relatively more frequently. However, a systematic review showed that serious adverse events are also underreported . Thus, we believe that serious adverse events, such as death, non-recovery, and sequelae, are common.
By calculating the ROR, we found that the most adverse events of CAM were infections and infestations, blood and lymphatic system disorders, hepatobiliary disorders, and immune system disorders. A study investigated the adverse events of switching from CAM to azithromycin (AZM) for MAC-LD when CAM was discontinued due to adverse events . Patients who switched from CAM to AZM did not have more adverse events, and the most common adverse event that led to discontinuation was skin rash. Kadota et al. investigated the effect of CAM-based regimens for MAC-LD in Japan . Adverse events occurred in 47.2% of CAM-based regimens, with 33.6% of adverse events related to CAM. The frequency of adverse events was not related to the total dose of CAM, doses per body weight, or age. The most common adverse events were skin rash, anorexia, hepatic dysfunction, diarrhea, and leukopenia. These results were not consistent with the results of this study because the regimens included drugs other than CAM, EB, and RFP. Kwon et al. compared the rate of treatment discontinuation due to adverse events of CAM and AZM for MAC-LD . The most common adverse event that resulted in treatment discontinuation was gastrointestinal disorders. The frequency of adverse events varies among studies because of the difficulty of studying only CAM. However, infections, infestations, and immune system disorders may be related to adverse events, respiratory symptoms, or immunodeficiency caused by MAC-LD. The spontaneous reporting system does not fully explain the causal relationship and may be subject to reverse causality bias. In this study, the onset of adverse events within 120 days was 40%. This suggests that attention should be paid to the occurrence of adverse events early in treatment. This study is novel in that we show that adverse events of CAM tend to occur early in the course of treatment.
In this study, the most common adverse events occurred between 181 and 300 days after the start of treatment, indicating that attention should be paid to the adverse events of EB, especially 6 months after the start of treatment. Adverse events of EB develop depending on doses, and risk factors include high doses per body weight and daily administration [22,23,24,25]. In a case series of 70 patients with EB-induced eye toxicity, 36 patients developed eye toxicity between 2 and 6 months, and 26 patients developed eye toxicity more than 6 months after the start of treatment . Therefore, the duration of treatment was also a risk. In a retrospective study of MAC-LD treatment in Japan, the median onset of eye disorders was as long as 278 days, which is a frequent cause of drug discontinuation (96.2%) . In a retrospective review comparing patients who received standard treatment, including EB, and those who discontinued EB due to adverse events, the treatment failure rate tended to be higher in the group of patients who discontinued EB . To avoid discontinuation of treatment, patients receiving EB should always be assessed for doses per body weight. In this study, 42.9% (36/84) of patients experienced non-recovery or sequelae, whereas 0.0% died. This may reflect the current situation in which patients are more aware of the adverse events of EB, are more regularly assessed for adverse events, and are more promptly treated when adverse events occur. In addition, using the Weibull distribution, we found that the onset time of eye disorder in EB was the wear-out failure type. This indicates that the incidence of adverse events increases over time. We should keep in mind that the treatment of MAC-LD requires long-term administration of EB and that the frequency of adverse events increases with the duration of treatment.
Using the ROR, we found that the most frequent adverse events when the suspect drug was RFP alone were renal and urinary disorders. In this study, we also found that most of the adverse events of RFP occurred within 120 days of administration. Few studies have examined the onset of adverse events of RFP alone, and we consider the results of this study to be important. Covic et al. investigated 60 cases of acute renal failure after RFP administration . The amount that induced renal failure and the total dose were not related to the severity of renal failure, and dialysis was directly correlated with the length of anuria. The common histological finding was tubulointerstitial nephritis . Muthukumar et al. reported 25 cases of acute renal failure due to RFP . Symptoms in 24 patients developed between 1 to 24 h, which are consistent with this study’s result. Among 12 patients who underwent kidney biopsy, 7 patients had acute interstitial nephritis. In addition, 1 patient had crescentic glomerulonephritis, and 3 patients had mesangial proliferation. Another review of 48 RFP-associated renal failure cases also showed the most common histological finding was acute tubular necrosis . Kim et al. reported a case who developed minimal change disease while taking RFP . These findings revealed that RFP could cause various types of kidney injuries. Moussa et al. compared the blood concentration of RFP in patients with tuberculosis who experienced adverse events , and there was no difference in blood concentration. Since there may be little relationship between the dose and occurrence of adverse events, we should note decreased urine output and anuria regardless of the dose of RFP.
In this study, a comprehensive review of adverse event reports in the treatment of NTM allowed us to improve identifying adverse events that are likely to be caused by the drug. For example, we identified renal and urinary disorders associated with RFP, which will guide us to be attuned to these disorders during treatment. Since the study was negative on the usefulness of RFP for NTM , it can be used as an indicator for changing treatment when an adverse event occurs. We found that adverse events caused by CAM and RFP tended to occur within 120 days of treatment initiation, whereas those caused by EB tended to occur after 180 days. Monitoring the patient at specific times, depending on the drug, may reduce the need for discontinuation and, therefore, the occurrence of treatment failure. In addition, the use of the JADER database made it possible to determine the situation in Japan in general and to collect a large number of cases compared with previous studies that summarized adverse events.
This study has some limitations. First, spontaneous reporting systems do not contain data on the total number of patients that use a drug. Therefore, signal detection by disproportionality analysis frequently focuses on differences in the ratio of adverse event reports . These results may be affected by unverified and incomplete reports, underreporting, and indication and reverse causality biases. Small sample sizes may increase these biases. Second, a previous study showed that the quality of reports differed depending on the facilities or occupations in the JADER database , and the quality of the reports was not evenly distributed. Third, although ROR is suggested to be less biased than the proportional reporting ratio, which is a ratio of reporting proportions, there are still various biases . We need to accumulate more cases to understand the actual situation of adverse events. Other data mining methods include the Gamma Poisson Shrinker program and Bayesian approaches such as the Bayesian Confidence Propagation Neural Network, which were not used in this study . Fourth, some reports did not describe specific outcomes, and there was a discrepancy between the number of adverse events and outcomes. In addition, multiple events may have been reported more than once at the same lesions. Fifth, there are several interaction signals [37, 38], but since MAC-LD treatment is in principle a three-drug combination, signals specific to interaction cannot be detected in this study. Sixth, the number of days of adverse events of EB was reported in 28 of 62 patients. Consequently, the analysis of the Weibull model may be inaccurate because of the large number of missing values. Seventh, we identified the most frequently reported sex and age for each drug. However, since we did not analyze non-cases, we could not perform a statistical analysis to determine if sex and age were risk factors. We need to accumulate more cases to understand the actual situation of adverse events.