The Prognostic Value of Serology in Persistent Q Fever Infection

Introduction

Q fever, caused by Coxiella burnetii, is a rare disease with significant consequences for patients with persistent localized infections. After primary infection, 1–5% of patients develop a persistent localized infection, formerly defined as chronic Q fever. While diagnosis and imaging of the disease have evolved during the last few decades, therapy is unchanged (Eldin et al., 2017).

Persistent infection requires prolonged antimicrobial therapy. A combination of doxycycline with hydroxychloroquine is first-line therapy, for a minimum of 18–24 months. Length of treatment is controversial. A fourfold decrease in phase 1 IgG (IgG1) and disappearance of phase 2 IgM (IgM2) at 1 year are considered good prognostic factors that should be reached with therapy (Anderson et al., 2013; Million and Raoult, 2015).

According to Million et al., beside clinical factors, the absence of a four-fold decrease in IgG1 and IgA at 1 year of follow-up was associated with increased mortality in Q fever endocarditis. Male sex, higher IgG1 at diagnosis, and delay in hydroxychloroquine treatment initiation were correlated with serological failure (Million et al., 2010).

In contrast, in 2021, Buijs et al. performed a large retrospective cohort study of 337 patients, and concluded that the course of IgG1 titers was not associated with first disease-related event or therapy failure. Poor prognostic factors were positive serum PCR during therapy and a higher IgG1 at the start of therapy (Buijs et al., 2021).

In our experience, many patients recover clinically after a few weeks or months of therapy, but serological response can be delayed for >24 months. Thus, we aimed to study the prognostic value of serology in our cohort of patients with persistent Q fever.

Methods

Results

During the study period 22 patients fulfilled diagnostic criteria for persistent Q fever, had full available data and follow-up, and were included in our cohort.

Demographic and clinical data are presented in Table 1.

We identified 17 patients with definite and 5 with possible diagnoses by the French criteria (Raoult, 2012) and 13 patients with proven and 9 with probable diagnoses by the Dutch criteria (Wegdam-Blans et al., 2012).

Ten patients had symptoms longer than 6 months before diagnosis.

Six patients (27%) were operated as part of the initial treatment. All of them had clinical cure, and 3 also had serological cure at EOT.

Serological and clinical outcomes are described in Table 2. Four patients died during follow-up––two had malignant diseases (one metastatic transitional cell carcinoma of the bladder, the other colon cancer), one from heart failure and one from an unknown cause––the latter two were considered related to Q fever for the purpose of the analysis. Fourteen patients had dermatological side effects (hyperpigmentation or a psoriatic like rash).

IgG1 and IgM2 titers during and after therapy are presented in Table 2 in detail.

Eighteen patients (82%) had clinical cure, and 10 out of 20 (50%) patients with complete available data had serological cure at EOT. One patient was referred for a valve replacement due to relapse after discontinuing therapy. One patient had a clinical and serological failure after a short course of antimicrobial therapy (7 months) since he refused to continue therapy despite deterioration of clinical condition. Detailed information of the patients that failed therapy is presented in Table 3.

Table 4 presents clinical and serological data of patients with clinical failure and clinical cure.

Patients who failed were slightly older than those who did not fail (72.8 ± 10 years vs. 60.9 ± 12 years respectively, p = 0.097). Treatment duration was similar in both groups (median 28.5 months for those who were cured and 24 for those who failed, p = 0.30), as other clinical characteristics.

IgG1 levels at presentation were significantly higher in the treatment failure group (median 9600 vs. 3200 in those who were cured, p = 0.019). At 1 year of treatment the values did not differ significantly (median 6400 in the failure group vs. 4800 in the cure group, p = 0.14), as in at EOT (3200 in the failure group vs. 1600 in the clinical cure group, p = 0.21). IgG1 was statistically significantly higher in the failure group than in the clinical cure group 6–12 months after EOT (median of 6400 vs. 800, p = 0.031).

In the failure group, IgM2 was positive at diagnosis for half of the patients and remained positive in two out of three patients at 1 year of therapy, compared with only one patient out of 15 (7%) who had clinical cure (p = 0.038). At EOT, one patient in each group still had positive IgM2.

Nine patients (53%) were defined as serological cure at the EOT in the clinical cure group versus one patient (33%) in the clinical failure (p = 1.0).

Discussion

In a cohort of patients with persistent Q fever, we identified 18 patients who had clinical cure and 4 who failed. We found that a higher IgG1 at presentation (median 9600 vs. 3200, p = 0.019) and at 6–12 months after EOT (median 6400 vs. 800, p = 0.031) and positive IgM2 after a year of therapy (67% vs. 7%) correlated with clinical failure.

Serological cure as previously defined (a four-fold decrease of IgG1 or IgG1 ≤ 800) was similar in both groups (33% vs. 53%, p = 1.0).

Clinical course of symptoms until diagnosis and treatment, choice of antimicrobial regimen, duration of antimicrobial therapy, and clinical and demographic risk factors were not significantly different between the groups.

The patients that were operated on early in the course of the disease did not have clinical failure, although due to small numbers this was not statistically significant.

Most of the patients in our cohort had prolonged treatment duration, with at least 18 months for native valves and 24 months for prosthetic valves, with a trend toward prolonged therapy if serology remained elevated.

Prolonged combination antimicrobial therapy is essential in patients with persistent Q fever, and is based on data from Marseille, France, which showed first that 3 years of therapy were needed to clear the valves from infection (Levy et al., 1991). Later, with combination therapy, a shorter duration of 18 months for native valves and 24 months for prosthetic valves was deemed as sufficient, as long as IgG1 levels dropped to ≤800, or at least showed a four-fold decrease. A positive IgM2 at 1 year or high IgG1 at 1 year, were considered poor prognostic factors (Million et al., 2010; Raoult et al., 1999). They, like us, described titers in defined time periods and not continuously, and this could miss other time periods that could be significant when addressing the best time to stop therapy.

Buijs et al. challenged this paradigm. In a retrospective cohort of 337 patients, 205 of them received treatment and follow-up of more than a year, IgG1 did not correlate with disease-related mortality or failure. The authors suggest that treatment decisions should be based on clinical response, imaging, and PCR (Buijs et al., 2021). It should be noted that median time to Q fever-related deaths was 55 weeks, which means that most of the complications in their study occurred during therapy, and not after EOT. Like in our study, almost 50% failed to reach serological cure at 1 year. They did find that positive PCR and high IgG1 at diagnosis correlated with therapy failure (Buijs et al., 2021).

Persistent Q fever patients differ in France and the Netherlands, with mostly endocarditis in France and more vascular infections in the Netherlands. These infections are not the same clinically and respond differently to antimicrobial therapy (Eldin et al., 2017). It is not surprising that results and conclusions from these two countries are very different. Our patients are more diverse, but in our cohort more patients had valvular risk factors than vascular risk factors; this could explain some of our findings.

In our study, only 1/22 patients showed a serological response to treatment at 1 year, but most showed a full clinical response. It might be that serology after 1 year of therapy is premature as a solid predictor for clinical cure.

We did not find a difference in outcome among patients who were treated with doxycycline and hydroxychloroquine versus doxycycline and ciprofloxacin, although the latter group was small. Van Roeden et al. found that these two regimens were similar in efficacy, with a median treatment duration of 95 weeks (van Roeden et al., 2018).

Follow-up after EOT is critical, since relapse is an important issue, especially during the first 5 years of follow-up, as described by Million et al. (Million et al., 2010). Our findings support this recommendation, with a high IgG1 6–12 months after EOT in our cohort, associated with clinical failure.

Higher IgG1 levels at presentation are known to be a poor prognostic factor (Million et al., 2010) that should suggest a closer monitoring and follow-up, and a case-by-case consideration of prolonged treatment.

Half of the patients who failed treatment (2/4) refused prolonged treatment. One patient was an 85-year-old female that had worsening heart failure with therapy, and gastrointestinal side effects. She died a few months after stopping therapy. Another patient was a 72-year-old male with a late diagnosis and poor compliance to therapy who refused to continue therapy after 7 months. He deteriorated slowly with fever, diffuse lymphadenopathy, weakness, high IgG1, without any evidence of malignancy.

The remaining two patients who failed treatment apparently followed our treatment recommendation. One patient is a 60-year-old male with prosthetic valves and good compliance to therapy, who is still under treatment, and the other was a 74-year-old female with a valvular and vascular disease, and prolonged therapy with good clinical recovery, who died a few years after EOT, with no specific details about the cause of death. Two patients who completed their antimicrobial therapy with good recovery, died from solid tumors. It should be mentioned that Q fever infection was associated with an increased risk for lymphoproliferative diseases, possibly due to overproduction of Il-10 (Melenotte et al., 2018; Melenotte et al., 2016).

None of our patients was diagnosed with lymphoma, and we cannot correlate the malignancies in two patients in our cohort to Q fever.

Our study has some limitations. First, it is a small-size cohort over two medical centers in Israel. Previous data from Israel included 35 patients between 1983 and 1992 (Siegman-Igra et al., 1997), since then no data were published in English literature. Our cohort, although small, had comprehensive and full medical records, with a long median follow-up. We think that the data we present is truly representative and can enrich us about persistent Q fever treatment.

Second, drug levels were not routinely available in Israel during our study period. Previous studies have correlated doxycycline levels with serological and clinical response (Rolain et al., 2005; van Roeden et al., 2018), and we are lacking that data.

Additionally, we did not check blood PCR in any of our patients, a marker suggested by Buijs et al. to be important in prognosis of persistent Q fever patients (Buijs et al., 2021). In their study, positive PCR was correlated with first disease-related events (HR 1.57, 95% CI: 1.10–2.25, p = 0.01); still most of these events occurred during first year of therapy, and we analyzed failures after EOT.

The results of the study should be interpreted with caution due to small numbers, p-value that could lead to wrong conclusions since with these numbers type-1 error is a possibility. We did not have a time-to-event analysis and could not stratify the patients according to length of treatment, surgical procedures etc. On the other hand, the strength of our study is driven by in-depth data about patients with valvular and vascular risk factors that were collected prospectively, with prolonged follow-up during and after EOT. The data, detailed in the tables and in the text, can assist clinicians when treating their own patients and coming to their own conclusions in this difficult-to-treat disease.

In conclusion, high IgG1 levels at presentation and at 6–12 months after EOT, and positive IgM2 after 1 year of therapy were found in our study as poor prognostic factors in persistent Q fever patients.

We do think treatment should be prolonged, especially in patients with high phase 1 IgG levels at presentation, since this is correlated with clinical failure. Serology does have a predictive role, but perhaps a greater role after EOT than during treatment, as raising levels of IgG1 at follow-up may indicate the need for restarting antibiotic therapy.

Close clinical and serological monitoring of patients, especially during the first few years after EOT, is essential, to recognize relapse early and treat properly. We suggest that a case-by-case approach should be made when clinical and therapeutic decisions are made, since Coxiella burnetii-persistent infection remains a great therapeutic challenge.