Improved penicillin susceptibility of Streptococcus pneumoniae and increased penicillin consumption in Japan, 2013–18

Objectives To examine the association between penicillin susceptibility of Streptococcus pneumoniae and penicillin consumption in Japan. Methods We used Japan Nosocomial Infection Surveillance data on the susceptibility of S. pneumoniae and sales data obtained from IQVIA Services Japan K.K. for penicillin consumption. We analysed both sets of data by decomposing them into seasonality and chronological trend components. The cross-correlation function was checked using Spearman’s rank correlation coefficient to examine the correlation between susceptibility and consumption. Results After adjusting for seasonality, the susceptibility of S. pneumoniae to penicillins gradually improved (55.7% in 2013 and 60.6% in 2018, respectively) and penicillin consumption increased during the same period (0.76 defined daily doses per 1,000 inhabitants per day [DID] in 2013, and 0.89 DID in 2018). The results showed positive cross-correlation (coefficient 0.801, p-value < 0.001). In contrast, cephalosporin consumption decreased (3.91 DID in 2013 and 3.19 DID in 2018) and showed negative cross-correlation with susceptibility of S. pneumoniae to penicillins (coefficient −0.981, p-value < 0.001). Conclusions The rates of penicillin-susceptible S. pneumoniae isolates did not negatively correlate with penicillin consumption at the population level. Increased penicillin consumption might not impair the penicillin susceptibility of S. pneumoniae.

After publication of the Global Action Plan on Antimicrobial Resistance in 2015 [1], the Japanese government established its National Action Plan on Antimicrobial Resistance the following year [11]. This plan includes S. pneumoniae as one of the major target organisms and aims to improve the microbe's susceptibility to penicillin. In this plan, the susceptibility of S. pneumoniae to penicillins is described in terms of the minimum inhibitory concentration (MIC) threshold for pneumococcal meningitis defined by the Clinical Laboratory Standards Institute (CLSI) in 2015 [12]. According to this criterion, 48% of S. pneumoniae isolates in Japan were resistant to penicillins in 2014, and the Ministry of Health, Labour and Welfare (MHLW) of Japan set a goal to reduce the proportion of penicillin-resistant S. pneumoniae (PRSP) to below 15% by 2020 [11].
Because the guidelines published by the CLSI are now used in Japan, we used the interpretive MIC breakpoints for meningeal and non-meningeal isolates of S. pneumoniae according to the 2008 revision of CLSI guidelines [13]. However, these figures change considerably when the non-meningitis MIC threshold is applied. In 2014, 99.6% of S. pneumoniae isolates were susceptible to penicillins and in 2018, the number rose slightly to 99.7% [14]. Under these circumstances, it is difficult for us to find a substantial difference in the proportion of susceptible isolates before and after the action plan. If we used non-meningitis MIC, it would be difficult to find a time-series trend because most S. pneumoniae would be classified as "sensitive". However, the sample size would be too small if we used only cerebrospinal fluid specimens to monitor the penicillin susceptibility of S. pneumoniae. Considering this, it might be reasonable to apply the meningitis MIC threshold to all types of specimens (sputum, cerebrospinal fluid, etc.) to follow the trend of AMR in S. pneumoniae.
Nevertheless, evaluation by meningitis MIC threshold is not an established method for surveying trends in the penicillin susceptibility of S. pneumoniae. For instance, Torumkuney and colleagues evaluated the susceptibility of S. pneumoniae from respiratory tract specimens via both non-meningitis and meningitis MICs [3,15]. Karlowsky and colleagues reported MIC 90 as an indicator of resistance [6]. However, because MIC 90 to penicillins was 0.12 mg/ L in most years, it is not a good alternative for monitoring the trend. Although the European Committee on Antimicrobial Susceptibility Testing (EUCAST) uses 0.06 mg/L as the threshold for penicillin susceptibility, it defined the threshold for penicillin resistance as 2.0 mg/L and did not mention the classification of isolates whose MICs were between 0.06 mg/L and 2.0 mg/L [16]. Additionally, EUCAST redefined its susceptibility testing categories (i.e. S, I, and R) in 2019 [17]. Under the new definition, S. pneumoniae isolates, which have MICs of between 0.06 mg/L and 2.0 mg/L, are evaluated as "Susceptible, increased exposure". Consequently, it is difficult to evaluate the penicillin susceptibility of S. pneumoniae according to EUCAST criteria only.
The Survey of Antibiotic Resistance has provided findings about AMR in various countries and reported the distribution among MICs including MIC50 and MIC90 [3,15,[18][19][20][21][22][23][24][25]. According to these previous studies, S. pneumoniae seldom showed MICs greater than 2 mg/L to penicillins in most countries, although the distribution of MIC values differed in each country. In other words, meningitis MIC might be a more appropriate threshold for monitoring the antibiotic susceptibility of S. pneumoniae. However, non-meningitis MIC might be beneficial for clinical use because most community-acquired pneumococcal pneumonia can be treated with penicillin regardless of the in vitro MIC value [26,27]. Another concern is the recommendation of antibiotic prescriptions for S. pneumoniae. Needless to say, antibiotics should never be prescribed unless they are necessary, because the use of any type of antibiotic could exert selective pressure that could lead to AMR [28,29]. Many guidelines recommend the penicillin class of antibiotics as the first choice for S. pneumoniae infections due to its effectiveness and narrow spectrum [30][31][32]. However, it is unclear whether increased penicillin consumption at the population level could lead to reduced penicillin susceptibility of S. pneumoniae.

PLOS ONE
This study had two main objectives. The first was to assess the appropriateness of meningitis MIC (0.06 mg/L) as an indicator for the penicillin susceptibility of S. pneumoniae. The other was to evaluate the relationship between penicillin consumption at the population level and the penicillin susceptibility of S. pneumoniae.

Data source
We used data collected by Japan Nosocomial Infections Surveillance (JANIS), which was organized by the Ministry of Health, Labour and Welfare [33,34]. The JANIS Clinical Laboratory module comprehensively collects all routine microbiological test results, including culturepositive and -negative results from over 2000 hospitals voluntarily participating in the surveillance, which account for a quarter of the approximately 8000 hospitals across Japan.
We extracted the monthly data for S. pneumoniae (including isolates from blood, cerebrospinal fluid, and respiratory tract specimens) and its susceptibility to antibiotics from January 2013 to December 2018 from the database of the JANIS Clinical Laboratory module. Patients were de-identified by each hospital before the data were submitted to JANIS. We included 636 facilities that continuously submitted their data to JANIS between 2013 and 2018. Approval for the extraction and use of the data was granted by the Ministry of Health, Labour and Welfare (0214-3).
To assess antibiotic consumption, we used monthly sales data collected by IQVIA Services Japan, which covers more than 99% of drug distribution among wholesalers in Japan. We calculated the monthly AMU using defined daily doses per 1,000 inhabitants per day (DID) from January 2013 to December 2018. We defined antimicrobials as J01 according to the ATC classification [35].

Statistical analysis
We prepared monthly time-series data about the rate of penicillin-susceptible S. pneumoniae isolates and sales volume of antibiotics by class (penicillins, cephalosporins, and sum of all classes). First, we decomposed both datasets into a trend component and a seasonality component to evaluate the chronological trend by using locally-weighted scatterplot smoother (LOWESS) method. Next, we used Spearman's rank correlation test to examine the correlation between components. Additionally, we checked the correlation between susceptibility of S. pneumoniae to penicillins and antibiotic consumption. A p value < 0.05 was considered statistically significant. All analyses were conducted by R version 3.6.3 [36].

Penicillin susceptibility of S. pneumoniae isolates
By applying non-meningitis MICs, the rates of penicillin-susceptible isolates were 97.8% in 2013 and 98.3% in 2018 (Fig 1). However, when applying meningitis MICs, the rates were 55.7% for 2013 and 60.6% for 2018 (Fig 2). The difference in the rates of susceptible isolates between the two cases is shown in Table 1 and   Table 2 and Figs 5-8.

Correlation between susceptibility and antibiotic consumption
The sales volume of penicillins positively correlated with the rate of penicillin-susceptible isolates (Spearman's correlation coefficient: 0.801, p < 0.001). The sales volume of cephalosporins and total antibiotics negatively correlated with susceptibility (Spearman's correlation coefficient: -0.981 and -0.888, both p < 0.001). The details of the correlation tests are presented in Table 3.

PLOS ONE
Susceptibility of S. pneumoniae and penicillin consumption

Discussion
In this study, we evaluated CLSI meningitis MIC (or EUCAST criterion) as an indicator for the chronological trend of penicillin susceptibility of S. pneumoniae. Our results showed that meningitis MIC can provide a more intelligible way of penicillin susceptibility follow-up than that of non-meningitis MIC. As mentioned above, JANIS uses MIC values defined by CLSI,

PLOS ONE
Susceptibility of S. pneumoniae and penicillin consumption and non-meningitis MICs do not reflect the present results because most of the S. pneumoniae isolates were penicillin susceptible according to non-meningitis MICs. We have already recognized the utility of CLSI non-meningitis MICs as a threshold for the clinical choice of antibiotics for preventing antibiotic abuse because most bacterial pneumoniae cases caused by S. pneumoniae can be treated by penicillin, even if the MICs are above 0.06 mg/L [26,27,[37][38][39]. Nevertheless, meningitis MICs are beneficial for AMR surveillance; therefore, the rate of penicillin susceptibility of all S. pneumoniae isolates should be assessed

PLOS ONE
Susceptibility of S. pneumoniae and penicillin consumption continuously by both non-meningitis and meningitis MICs. EUCAST criteria are useful if we regard 0.06 mg/L as the threshold for surveillance and 2.0 mg/L as the threshold for clinical use.
As for trend of antibiotic use, abuse of unnecessary antibiotics has been an important health issue in Japan, then various interventions have been implemented in Japanese healthcare system. For example, Ministry of Health, Labour and Welfare Japan became to provide incentives to healthcare facilities which implemented infection control team and other interventions in

PLOS ONE
Susceptibility of S. pneumoniae and penicillin consumption 2012 [40]. Although publication of National Action Plan was 2016, such interventions were considered to have influence on antimicrobial consumption at national level.
Increase of penicillin consumption can be considered as a result of promoting appropriate antibiotic use because cephalosporin and other classes of antibiotic consumption in Japan was higher than that in other countries [41]. When we reduced inappropriate cephalosporin and other classes of antibiotic consumption, a part of it should be replaced by penicillin.
Another novel finding is that the increase in penicillin consumption is not negatively associated with an increase in penicillin susceptibility in S. pneumoniae. On the contrary, our results showed that the sales volume of penicillins positively correlated with the rate of penicillin-susceptible isolates.
Of course, this should not be interpreted directly because the National Action Plan on Antimicrobial Resistance encouraged not only the use of narrow-spectrum antibiotics like penicillins but also the reduction of unnecessary prescriptions of any class of antibiotics [11]. As a result, the sales of cephalosporins and the sum of all classes of antibiotics decreased gradually during the study period. Similarly, antibiotic use decreased during the same period [42,43]. The improved penicillin susceptibility rate of S. pneumoniae might be explained by this gradual decrease in total antibiotic consumption; hence, the positive correlation between the consumption of penicillins and susceptibility is not a causative relationship. Indeed, the negative correlation between total antibiotic consumption and penicillin susceptibility of S. pneumoniae is compatible with the findings of previous studies [44,45].
Considering these findings, it seems that we do not need to change our current recommendation, which is the use of penicillin as the first choice for the treatment of pneumococcal pneumonias and bloodstream infections, because it will be unlikely to induce penicillin resistance, and the use of total antibiotics and cephalosporins is more strongly associated with a decrease of penicillin-susceptible isolates. It is possible that even in case the increase of penicillin consumption negatively influences the susceptibility of S. pneumoniae, its influence might be cancelled by positive influence of consumption reduction of other classes. To encourage use "Access" drugs would be desirable in order to optimize antibiotic use at the population level [46][47][48].
Our data demonstrated obvious seasonality. Both sales data and penicillin resistance peaked in winter and were at their lowest levels in summer, similar to previous studies [44,45,[49][50][51]. The correlation between the sales volume and susceptibility was strongest without lags, rather than with lags. Therefore, we are not sure about the length of time needed to observe the influence of changes in sales of antibiotics on susceptibility. It is easy to understand why antibiotic consumption peaks in winter, which is because acute respiratory tract infections are more commonly seen during this season; therefore, both appropriate and inappropriate use of antibiotics increases in winter. However, we are not sure why antibiotic resistance in S. pneumoniae peaks at the same time. This is another issue for further consideration.

Susceptibility of S. pneumoniae and penicillin consumption
This study had several limitations. First, our results showed correlation rather than causation. It is likely that increased penicillin consumption does not induce penicillin resistance in S. pneumoniae at the population level; however, it is also possible that improved penicillin susceptibility induces an increase in penicillin prescriptions. Second, we used only sales data and susceptibility data, and other factors (e.g. genetic mechanism) were not included in our analyses. Because various interventions, which include education for the general population, were implemented after (and even before) publishing the National Action Plan on Antimicrobial Resistance in Japan, social factors other than the decrease in antibiotic consumption might have had an influence on the susceptibility of S. pneumoniae [52][53][54][55]. If we can take other societal factors into consideration, we may be able to evaluate the influence of change in antibiotic consumption on susceptibility more precisely.

Conclusions
Our findings suggest that meningitis MICs defined by CLSI might be useful indicators for the continuous monitoring of the susceptibility of S. pneumoniae to penicillins. The increase in penicillin use is not positively associated with susceptibility of S. pneumoniae isolates, which suggests that the current recommendations are reasonable. Further study including genetic mechanism of penicillin susceptibility of S. pneumoniae would reveal additional details of the relationship between antibiotic consumption and antibiotic resistance in S. pneumoniae.