Health status and concomitant prescription of immunosuppressants are risk factors for hydroxychloroquine non-adherence in systemic lupus patients with prolonged inactive disease.

Background/objective The objectives of this paper are to assess the extent of and the factors associated with hydroxychloroquine (HCQ) non-adherence in systemic lupus erythematosus (SLE) patients with prolonged inactive disease and to investigate relationships between blood HCQ concentration and quality of life (QoL). Methods Consecutive SLE patients, in remission for at least one year and taking a stable dose of HCQ were investigated. At study entry (T0) and six months later (T6) a blood venous sample was taken to measure whole blood concentration of [HCQ] and desethylchloroquine ([DCQ]). Moreover, at T0 each patient completed validated questionnaires assessing QoL, disability, anxiety, depression and visual analogue scales for fatigue, pain, general health (GH), and self-assessment of disease activity. Results Eighty-three patients with a median [HCQ] of 327 ng/ml were enrolled. At T0, 24 (29%) were defined as non-adherent ([HCQ] < 100 ng/ml). At multiple logistic regression analysis the physical summary of SF-36 (p = 0.038), and the concomitant use of immunosuppressants (p = 0.010) were independently associated with non-adherence. A significant increase of HCQ adherence was observed at T6 (p < 0.05). Conclusions A better health status and the concomitant prescription of immunosuppressants represent risk factors for HCQ non-adherence in SLE patients in remission. Monitoring HCQ levels might represent an important opportunity to improve adherence.

Systemic lupus erythematosus; hydroxychloroquine; quality of life; treatment adherence

Systemic lupus erythematosus (SLE) is an autoimmune systemic disease involving almost every organ/system[1] that still burdens patients with an impaired quality of life (QoL) and a higher mortality than the general population.[2]

Hydroxychloroquine (HCQ) represents a milestone in the treatment of SLE patients and to date its intake is recommended for a lifetime because of its known capacity to prevent flares and thromboembolic events, to improve lipid and glucose metabolism, and to reduce damage accrual over time.[3] Given its long elimination half-life, the assessment of HCQ concentration in whole blood is of increasing importance in clinical practice because it is a very specific measure of treatment adherence.[4] Moreover, low blood HCQ concentration is associated to a higher SLE disease activity,[5] and independently predicts SLE flares within a six-month period.[5] Therefore, measurement of HCQ levels in whole blood could help identify patients at risk who require a stricter follow-up.

Despite the increasing interest in the topic, the extent of poor adherence to HCQ treatment, its main originating factors and the relationship with disease progression have not been yet sufficiently investigated in SLE patients with prolonged inactive disease.

Moreover, data from cohort studies report that the QoL in SLE patients achieving a durable period of clinical remission is generally worse than that recorded in the general population because of the presence of aspecific symptoms such as malaise, fatigue, and musculoskeletal pain.[6] These symptoms can be explained at least in part by a residual subclinical disease activity, not captured by most of the available activity scores.[7] Since HCQ is known to have a beneficial role in pain control, skin manifestation and overall disease activity,[3] and given that the achievement of an optimized therapeutic drug concentration is associated with a better disease outcome,[5] we were interested in investigating the relationship between blood concentration of HCQ and QoL in such patients.

Therefore, we designed this study to estimate the extent of and the main demographic, clinical and laboratory factors associated with HCQ non-adherence, and the relationship between patient-reported outcomes and HCQ blood concentration in SLE patients with prolonged (>1 year) inactive disease.

The study participants included patients consecutively admitted to the Rheumatology Unit of the University of Campania “Luigi Vanvitelli” in Naples (previously named Second University of Naples) from 30 November 2014 to 30 April 2016, who agreed to take part in the study and were able to sign a written informed consent. Each patient had to satisfy the following inclusion criteria: fulfillment of the 2012 classification criteria of the Systemic Lupus International Collaborating Clinics (SLICC);[8] state of complete or clinical remission (with or without treatment) according to the preliminary Definitions of Remission in SLE (DORIS) criteria,[9] for at least one year; treatment with a stable dose of oral HCQ during the previous six months. If taken, immunosuppressants and glucocorticoids had to be prescribed at a stable dose during the previous month. All patients with concomitant fibromyalgia, with known psychiatric disorders or pregnancy were excluded from the study.

On admission, each patient underwent a detailed history taking, physical examination and laboratory investigations to assess disease activity by the Safety of Estrogens in Lupus Erythematosus National Assessment SLE Disease Activity Index (SELENA-SLEDAI),[10] and disease damage by the Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index of SLE.[11] Creatinine clearance was measured according to the Modification of Diet in Renal Disease (MDRD).[12] Remission was defined according to the DORIS proposal as follows: complete remission without treatment, i.e. no clinical (SLE-related) manifestations, no serologic abnormalities (low C3 and/or C4, increasing double-stranded DNA (dsDNA) titre), only antimalarials allowed; complete remission on treatment, i.e. no clinical manifestations, no serologic abnormalities, therapy with antimalarials, daily prednisone ≤5 mg and/or immunosuppressants allowed; clinical remission without treatment, i.e. no clinical manifestations, presence of serologic abnormalities, only antimalarials allowed; clinical remission on treatment, i.e. no clinical manifestations, presence of serologic abnormalities, therapy with antimalarials, daily prednisone ≤5 mg and/or immunosuppressants allowed.[9] Moreover, the following data were collected: height, weight, body mass index, time of the intake of the last tablet of HCQ, daily dose of HCQ, smoke status (current/past smoker). In addition, each patient was examined to assess the co-existence of fibromyalgia according to published diagnostic criteria,[13] and a careful interview about the intake of drugs for psychiatric problems was also performed.

Each patient was then asked to complete a visual analogue scale (VAS) for pain, fatigue and self-assessment of disease activity which ranged from 0 (none) to 100 mm (very severe/active); a VAS for global health (GH) status which ranged from 0 (poorest) to 100 mm (the best); the Italian version of Short-Form 36 (SF-36), summarized in two composite summary scores, the physical component summary score (PCS) and the mental component summary score (MCS), ranging from 0 (poorest QoL) to 100 (highest);[14] the Italian version of the Health Assessment Questionnaire Disability Index (HAQ-DI), ranging from 0 (no disability) to 3 (highest disability);[15] and The Hospital Anxiety and Depression Scale to assess the level of anxiety and depression, where a score of 0 to 7 for either subscale was considered to be in the normal range.[16] ,[17]

Finally, a venous blood sample was collected to measure blood concentration of HCQ and desethylchloroquine (DCQ), as described below. All the patients were unaware that HCQ and DCQ blood concentrations would have been measured until they attended the outpatient clinic, when they were asked to be enrolled in the study.

Treatment was prescribed according to latest published guidelines.[18] The prescribed daily HCQ dose was up to 6.5 mg/kg. If the number of daily-prescribed pills was more than one, patients were advised to take them twice a day.

A follow-up visit was planned for each patient after six months (T6). At T6, each patient underwent the same clinical and laboratory assessment and was asked to take another unscheduled venous blood sample to measure whole blood HCQ and DCQ concentration. The occurrence of a flare during the follow-up was assessed by the SELENA/SLEDAI flare composite score.[19]

A venous whole blood sample was collected in a tube containing 125 units of heparin and stored at −20℃. Heparinized whole blood, stored at −20℃, was thawed at room temperature. Five-hundred microlitres of whole blood were transferred to a 15 ml Falcon tube and added to 500 μl of deionized water and 300 μl of 25% ammonia solution (Molecular Biology grade, Applichem). Each sample was extensively vortexed, extracted with 4 ml of diethyl ether and centrifuged at 1200 g for 5 minutes at 4℃. The sample extracted was kept for 15 minutes at −80℃ to allow complete freezing of the aqueous phase. The liquid organic layer was then transferred to a clean tube and air-dried O/N. The residue was suspended with 300 μl of 5% acetonitrile and 0.1% formic acid and the resulting mixture was centrifuged at 19,000 g, 4℃ for 5 minutes. The supernatant was then transferred to a clean vial and injected (1 μl) in a UPLC-MS/MS system composed of a Nexera chromatograph (Shimadzu) coupled with a Q-trap 6500 spectrometer (AB Sciex). Chromatographic separation of HCQ and related metabolites was carried out using a Kinetex F5 Core-Shell LC Column (Phenomenex, 100 × 2 mm, 2.6 µm particle size) thermostated at 40℃. Elution was performed using 0.1% formic acid (Eluent A) and 0.1% formic acid in acetonitrile (Eluent B), and a 3-minute gradient ranging from 5% to 50% of Eluent B, with a flow rate of 250 µl/min. Mass analysis was performed in positive MRM mode, monitoring the following transitions: 336.8→247.3 HCQ; 292.8→179.6 DCQ. Data were analysed using “Analyst” software (AB-Sciex).

Non-adherence was defined as a whole blood [HCQ] < 100 ng/ml.[4] ,[20]

The study was approved by the Ethics Committee of the University of Campania “Luigi Vanvitelli”.

Continuous variables were analysed with the unpaired Student’s t-test or the Mann-Whitney test, or with the paired t-test and Wilcoxon test, where appropriate. The chi-square or Fisher’s exact test was applied for categorical variables. Spearman correlation was used to assess relationship between [HCQ] and continuous patient-related outcomes variables. P values less than 0.05 were considered significant. Univariate logistic regression analysis was constructed to assess factors associated with HCQ non-adherence, identifying as dependent variables an [HCQ] < 100 ng/ml. The factors found to be significant in univariate analysis (p < 0.1) were entered in a multivariate model. For continuous predictive variables, odds ratios (ORs) expressed the risk associated with a one standard deviation (SD) increase for each continuous predictive variable. Statistical analysis was performed with the MedCalc software, version 12.7.0.

During the study period, 83 consecutive SLE patients re-admitted to the Rheumatology Unit of the Second University of Naples fulfilled the inclusion criteria and were enrolled in the study after giving written informed consent. They were mostly female (95%), with a mean ± SD age of 41 ± 11 years, a median disease duration of 15 years (range 2–37 years), and a median SLICC damage index of 0 (range 0–3). Fifty-one (61%) were in remission with treatment, the remaining were in remission without treatment (on HCQ only). The mean ± SD dose of HCQ per weight prescribed was of 5.3 ± 1.2 mg/kg and the mean time from last pill intake was 9.8 ± 8.7 hours. The median [HCQ] at baseline was of 327 ng/ml (range 0–4003 ng/ml), that of [DCQ] of 47 ng/ml (range 0–650 ng/ml). There was no significant difference between patients in remission without or with treatment in both [HCQ] (median 285 ng/ml (range 0–1723 ng/ml) versus median 435 ng/ml (range 0–4003 ng/ml); p = 0.512) and [DCQ] (median 47 ng/ml (range 0–239 ng/ml) versus 51 ng/ml (range 0–650 ng/ml); p = 0.619).

Clinical and therapeutic features of the patients enrolled are summarized in [1] .

Clinical, demographic and laboratory features of 83 patients enrolleda

-1 If not otherwise specified, the values are the number (%) of patients. [HCQ]: whole blood concentration of hydroxychloroquine; [DCQ]: whole blood concentration of desethylchloroquine; SLICC: Systemic Lupus International Collaborating Clinics.

Among the 83 patients, 24 (29%) had undetectable blood [HCQ] (n = 17) or an [HCQ] < 100 ng/ml (n = 7, mean ± SD 67 ± 34, range 9–99), thus reflecting a very poor adherence,[4] ,[20] and additional four patients showed a 100 ≥ [HCQ] < 200 ng/ml. No patient had undetectable [DCQ], a condition suggesting a very recent resumption of treatment in patients with dosable HCQ.[21]

We found in only 11 (14%) patients an [HCQ] ≥ 1000 ng/ml, a value considered the threshold associated with the lowest probability of flare within six months in unselected SLE patients.[5]

At baseline, median physical component summary score (PCS) and the mental component summary score (MCS) were, respectively, 47 (range 12–62) and 45 (range 18–66). The self-reported median VAS were as follows: VAS pain 17 mm (range 0–100 mm), VAS fatigue 33 mm (range 0–100 mm), VAS GH 74 mm (range 20–100 mm), and VAS patient’s self-assessment of disease activity 20 mm (range 0–94 mm). The median HAQ-DI score was 0 (range 0–2.125). [2] shows the results of self-assessment parameters in the overall sample.

Patient’s related outcome measures in overall sample, in non-adherent and adherent patients

-2 [HCQ]: whole blood concentration of hydroxychloroquine; PCS: Physical Component Summary score; MCS: Mental Component Summary score; HAQ-DI: Health Assessment Questionnaire Disability Index; VAS: visual analogue scale; mm: millimetre; GH: global health status. Statistically significant comparisons are in bold (p < 0.05).

In the overall study population, a significant weak positive correlation was found between [HCQ] and both VAS pain (rho = 0.275; p = 0.012) and patient’s self-assessment of disease activity (rho = 0.267; p = 0.016). A negative correlation between [HCQ] and some domains of SF-36 such as limitations due to physical problems (expressed by the domain Role limits – Physical) (rho = –0.233; p = 0.04) and bodily pain (rho = –0.228; p = 0.04) was instead retrieved. Moreover, a trend towards a negative correlation was found between [HCQ] and the physical summary component score of SF-36 (PCS) (rho = –0.217; p = 0.056).

Since these results can be influenced by the lowest non-therapeutic [HCQ] found in non-adherent patients, we excluded such patients from the analysis. The analysis of the remaining patients failed to show any significant correlation between [HCQ] and each of the investigated parameters, both at study entry and at follow-up visit (data not shown).

At univariate analysis, poor-adherence patients reported a better QoL in physical domain (median PCS 50 (range 29–60) versus 44, range (12–62); p = 0.021), lower level of pain (median VAS pain 3 mm (range 0–55 mm) versus 25 mm (range 0–100 mm); p = 0.011), a higher reported GH (median VAS GH 80 mm (range 32–100 mm) versus 67 mm (range 20–100 mm); p = 0.043), a lower self-assessed disease activity (median VAS patient 5 mm (range 0–70 mm) versus 23 mm (range 0–94 mm); p = 0.028) and a trend towards a lower disability (median HAQ 0 (range 0–0.875) versus 0.625 (range 0–2.125); p = 0.062), compared to other patients. No significant difference was found either for the reported level of fatigue, anxiety, depression or for any available demographic or clinical features potentially influencing HCQ metabolism (HCQ prescribed dose/weight, body mass index, creatinine clearance, smoke) between the two groups ([1] ). Concerning the treatment, the number of non-adherent patients on concomitant immunosuppressants therapy was significantly higher than that of adherent patients (13/24 versus 10/59; p = 0.001), while there was no difference in the number of GC users (p = 0.974) ([1] ).

At multivariate logistic analysis, the factors independently associated with non-adherence were the physical summary of QoL (PCS) as assessed by SF-36 (OR 1.05, 95% confidence interval (CI) 1.00–1.11; p = 0.038) and the concomitant use of immunosuppressants (OR 4.35, 95% CI 1.41–13.44; p = 0.010).

Of the 83 enrolled patients, 77 (93%) attended the scheduled follow-up visit after mean ± SD 7 ± 2 months. Among them, five (6%) had a minor flare (four skin rash; one skin rash and low platelet count). No major flare was observed. The occurrence of flare was not found to be associated with any patient’s clinical or demographic features, reported outcome measures, low dsDNA positivity and/or hypocomplementemia (data not shown). The median baseline [HCQ] of patients who flared was of 284 ng/ml (range 0–1723 ng/ml); for patients who did not flare the value was of 435 ng/ml (range 0–4003 ng/ml) (p = 0.225). Fifty-four among the 77 patients (70%) (14/20 non-adherent, 40/57 good adherent; p = 0.787) gave their consent to have another unscheduled blood sample to measure blood HCQ.

Patients labelled as non-adherent at study entry achieved at T6 a significant higher [HCQ] than baseline (median 0 ng/ml (range 0–99.4 ng/ml) versus median 515 ng/ml (range 0–1682 ng/ml); p < 0.0001), comparable to what recorded in better adherent patients (median 515 ng/ml (range 0–1682 ng/ml) versus 631 ng/ml (range 0–2208 ng/ml); p = 0.385). Only two out of 13 patients with [HCQ] ranging from 100 to 300 ng/ml at T6 had undetectable [DCQ], indicating a recent resumption of treatment.

In this study, we have investigated the prevalence of non-adherence as assessed by blood HCQ concentration measurement and the factors associated with, including the main QoL self-reported outcomes, in SLE patients with inactive disease. We pointed out that at least one-third of inactive SLE patients did not take the drug properly, and contrary to what we assumed, this happened mostly in those with better QoL, lower level of pain, lower self-rated disease activity, and in those with a concomitant use of immunosuppressants. The number of observed flares within a six-month follow-up was low and occurred in patients with low blood concentration of HCQ. We confirmed also that the patient’s awareness of having a blood measurement of HCQ might have led to an improved adherence.

Poor adherence represents an important cause of treatment failure in patients affected by chronic diseases.[22] Unfortunately, establishing the burden of non-adherence is not easy in clinical practice, and the self-reported questionnaires have shown to be mostly unreliable.[23] However, clinicians taking care of SLE patients on chronic HCQ can benefit from the use of blood monitoring, which is a very specific tool to estimate drug adherence.[4]

Previous studies conducted on SLE patients have shown that the rate of non-adherence assessed by HCQ blood measurement is highly variable,[4] ,[24] ranging from 7% up to 30% when the analysis is restricted to patients with active disease.[4] We chose a low threshold to define non-adherence (i.e. [HCQ] < 100 ng/ml)[4] ,[20] to exclude any possible interference with factors known to influence HCQ blood concentration, and found a high rate of HCQ non-adherence, similar to what reported for patients with active SLE in other surveys.[4] Moreover, we observed a very low proportion of patients (about 10%) reaching an [HCQ] ≥ 1000 ng/ml, which is considered the threshold associated with the lowest probability of flare within six months in unselected SLE patients.[5]

Despite these findings, the number of relapses observed by us was comparable to that described by Costedoat-Chalumeau et al. (6% vs 12%; p = 0.340) in their work, where low HCQ was the only predictor of SLE flares in a multivariate analysis.[5] In that specific survey, [HCQ] values were generally significantly higher than those observed by our group (mean [HCQ] of 1017 ng/ml). The discrepancy between a high rate of poor adherence and the lower than expected number of flares could be explained by the differences in concomitant treatment (i.e. corticosteroids, immunosuppressants) between the two cohorts, and/or by inclusion in our study of patients with prolonged inactive disease and, thus, less likely to relapse. However, it should be also kept in mind that our study was unpowered to investigate the relationship between HCQ intake and disease course, because it was not designed to primarily answer this question. In this context, we could only speculate that the target blood concentration (if any) required to prevent flares in this subgroup of patients could be lower than the value of 1000 ng/ml reported for unselected SLE patients.[5] The identification of such a lower threshold in larger studies with a longer follow-up would be very useful to avoid over-treatment, also according to the American Academy of Ophthalmology, for which the maximum recommended daily dosage of HCQ to minimize the risk of retinopathy has been lowered to <5.0 mg/kg.[25]

To date, the relationship between blood levels of HCQ and QoL of SLE patients has been investigated, to the best of our knowledge, only in a recent study.[26] The authors analysed data derived from patients enrolled in a randomized controlled trial (PLUS study) including only patients with a suboptimal blood concentration of HCQ (ranging from 100 to 750 ng/ml); the main objective of their study was to assess if adjusting the HCQ dose to reach the therapeutic threshold of 1000 ng/ml might lead to a lower rate of flares.[20] In their analysis, in which poor adherent patients were excluded from the study by definition, they found no association between blood levels of HCQ and any of the SF-36 domains.[26]

Our study shows some differences with that of Jolly and colleagues and adds further information on the topic. First, we primarily focused on patients with persistent inactive disease, in whom such kind of relationship, to the best of our knowledge, has not yet been investigated. Second, we excluded patients with fibromyalgia and psychiatric disorders to avoid the influence that these conditions have on physical and mental status of patients with connective tissue diseases.[27] Moreover, we did not limit the analysis to SF-36, but we explored relationship of HCQ blood concentrations with other important domains such as fatigue, pain, patient’s self-rated disease activity, anxiety and depression.

The key point of our work, which represents also the main difference with the above-mentioned study, is that we were able to appraise the relationship between patient’s QoL and non-adherence to HCQ. Contrary to what we had initially hypothesized, a better health status in physical domain was associated with a very poor adherence to HCQ intake. Actually, non-adherent patients were those rating their disease as less active, had a lower degree of pain and a better GH perception and QoL related to physical problems. This observation led us to hypothesize that intentional factors (i.e. related to a conscious patient’s choice) might represent the main contributors of an incorrect intake of HCQ. Probably, an acceptable health status could have prompted patients to lighten their drug intake. This hypothesis is in line with the findings of a French study[4] conducted on 209 SLE patients in which nearly all non-adherent patients reported that they had decided not to take HCQ because of concerns about potential side effects, the belief that HCQ was less effective than other treatments, or the decision that the disease itself did not deserve such treatment anymore. Additionally, our finding that patients taking immunosuppressants were more likely to miss HCQ pills intake might be due either to their belief that immunosuppressants alone are sufficient to control disease or to patient’ willingness to avoid potentially harmful drug overuse. However, as already mentioned, issues of unintentional non-adherence coming from the need to take many pills cannot be completely excluded.

As in other reports,[4] ,[20] ,[28] the rate of non-adherent patients decreased after enrolment in the study. In fact, at the second sampling only about 10% of patients, compared to one-third at baseline, had very low levels of blood HCQ and the HCQ concentration in patients labelled as non-adherers at study entry reached a value comparable to those with a better adherence at their follow-up visit. The improved adherence following the monitoring of the HCQ dose has already been described by other authors. Repeated HCQ measurements, patient counselling and mail reminder led to achieve good adherence to HCQ in 80% of patients studied by Durcan et al.[28] Treatment adherence significantly rose after discussion with treating physician also in the paper by Costedoat-Chalumeau et al.[4] Interestingly, we observed an improved adherence despite the fact that patients were not aware that a second measurement was performed and that they did not receive any information about the result of their previous blood sample. Therefore, the improved adherence might come from the mere patient’s suspicion to be more strictly monitored regarding his or her attitude to taking the drug, regardless of the knowledge of the results of a previous HCQ measurement or from the occurrence of an interview with the treating physician. The present observation might represent an additional advantage of routinely measuring HCQ blood concentration.

To recognize the key messages coming from this study, some further comments are required. Although the assessment of blood HCQ concentration is highly specific to capture non-adherent patients, it is necessary to keep in mind that its sensitivity is lower than specificity because the therapeutic concentration range can be reached only after a few days from starting drug intake.[29] Consequently, the quote of non-adherers can be notably higher than what we were able to identify through HCQ blood concentration measurement at one point. However, we hypothesize that the influence of a resumed intake of the drug just before the visit in patients included in our study was not meaningful because no patient was aware of this study before attending the visit, and the measurement of HCQ had never been routinely performed in our unit before the start of the present study. Moreover, we concomitantly measured the blood concentration of one of the main HCQ metabolites, the DCQ, to capture those patients reassuming the drug a few days before the visit. In fact, in such patients’ DCQ is generally undetectable despite the possibility to find acceptable concentration of HCQ.[21] We found only two patients with undetectable DCQ and suboptimal HCQ blood levels.

Another aspect to be considered is that the short follow-up period did not allow us to draw firm conclusions on the extent of poor adherence over time and its influence on health status and disease course. Moreover, the low number of events (flares) observed hampered us to identify clinical, serologic or patient’s reported outcomes associated with relapse and to understand whether the blood concentration of HCQ responsible for minimizing the risk of flares in patients with persistent inactive disease is different from that estimated in an unselected sample of SLE patients.[5]

In conclusion, we found that having a blood concentration of HCQ above the identified threshold of non-adherence was not associated with a lower extent of pain, fatigue, mood disorders and perceived disease activity in SLE patients with inactive disease. On the contrary, patients who felt better and those taking immunosuppressants were more prone not to take HCQ properly. Additionally, we pointed out that the mere suspicion of having a routine blood measurement of HCQ blood concentration effectively improved treatment adherence.

Our results confirm the utility of performing routine blood monitoring of HCQ and its metabolites concentration in clinical practice as a tool to identify non-adherent patients. The awareness of a quite frequent poor-adherent behaviour, which generally is not limited to a single drug or physician prescription, has to be taken into account when facing the occurrence of a disease flare or the management of comorbidities in such patients.

Studies with longer follow-up will help to better define the impact of a persistent (over years) correct intake of HCQ on important clinical outcomes.

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

The authors received no financial support for the research, authorship, and/or publication of this article.