Impact of lopinavir-ritonavir exposure in HIV-1 infected children and adolescents in Madrid, Spain during 2000-2014

Background The most-used protease-inhibitor in children is Lopinavir-ritonavir (LPV/r), which provides durable suppression of viral load and increases CD4+T-counts. This study describes the virological outcome of the HIV-1-infected paediatric population exposed to LPV/r during 15 years in Spain. Methodology Patients from the Madrid Cohort of HIV-1-infected-children and adolescents exposed to LPV/r as different line therapy during 2000–2014 were selected. The baseline epidemiological-clinical features, viral suppression, changes in CD4+T-CD8+T cell counts and drug susceptibility were recorded before and during LPV/r exposure. Drug resistance mutations (DRM) were identified in viruses from samples collected until 2011. We predicted drug susceptibility to 19 antiretrovirals among those carrying DRM using the Stanford′s HIVdb Algorithm. Results A total of 199 (37.3%) of the 534 patients from the cohort were exposed to LPV/r during 2000–2014 in first (group 1), second (group 2) or more line-therapies (group 3). Patients were mainly Spaniards (81.9%), perinatally infected (96.5%) with subtype-B (65.3%) and HIV-diagnosed before year 2000 (67.8%). The mean age at first LPV/r exposure was 9.7 years. After protease-inhibitor exposure, viral suppression was higher in groups 1 and 2 than in group 3. Viral suppression occurred in 87.5%, 68.6% and 64.8% patients from groups 1, 2 and 3, respectively. Among the 64 patients with available resistance data during LPV/r treatment, 27(42.3%) carried DRM to protease-inhibitor, 28 (58.3%) to reverse-transcriptase-inhibitors and 21 (43.7%) to non-reverse-transcriptase-inhibitors. Darunavir/ritonavir, atazanavir-ritonavir and tipranavir/ritonavir presented the highest susceptibility and nelfinavir the lowest. Conclusions A better lymphocyte recovering occurred when protease-inhibitor was taken as part of a first-line regimen and a higher number of patients reached viral suppression. The least compromised antiretrovirals for rescue antiretroviral regimens, according to DRM in the LPV/r-exposed-paediatric cohort, were mainly the new protease inhibitors.

particularly acknowledge the patients in this study for their participation and the Pediatric HIV BioBank integrated in the Spanish AIDS Research Network for providing two clinical samples used in this work. The Pediatric HIV BioBank, integrated in the Spanish AIDS Research Network, is supported by Instituto de Salud Carlos III, Spanish Health Ministry (Grant n˚RD06/0006/0035) and Fundación para la investigación y prevención del SIDA en España (FIPSE). CoRISPe cohort is funded by RD12/0017/0035 and RD12/0017/0037 Project, supported by Plan Nacional de I+D+I and cofinanced by the ISCIII-Subdirección General de Evaluación and European Development Regional Fund "A way to achieve Europe" (ERDF). We thank CoRISPE and the professionals responsible for the routine resistance testing in the participant hospitals for providing epidemiological-clinical data and pol sequences of enrolled patients when available.
changes in viraemia, CD4+T-CD8+ lymphocyte T cell counts and DRM comparing baseline data and the last available determination during LPV/r experience until December 2014. Patients exposed to LPV/r as off-label therapy were also identified. The Madrid Cohort of HIV-1-infected-children and adolescents, established in 2003, includes the 36% of the infected paediatric population registered in Spain before the end of December 2014.

HIV-1 pol sequencing
HIV-1 RNA extraction, amplification and sequencing were performed in pol coding region [protease (PR) and partial reverse transcriptase (RT), 1,121 bp] as previously reported [24]. New samples from 2 patients were provided by the Paediatric HIV BioBank integrated in the Spanish AIDS Research Network (RIS) RD12/0017/0035 and RD12/0017/0037 [25]. Samples were processed following current procedures and frozen immediately after their reception. All patients participating in the study gave their informed consent and protocols were approved by institutional ethical committees. The parents or legal guardians of all patients participating in the study gave their written informed consent to this BioBank, responsible for the storage of HIV-1 infected samples from the children under follow-up in Madrid used for this research. The project was approved by the Human Subjects Review Committee at Hospital Universitario Ramón y Cajal (Madrid, Spain).

Genotypic drug resistance identification
For the HIV drug resistance study, the patients were selected according to their pol sequence, genotypic resistance profile or sample availability by 2011. Most of these cases had been previously reported [24]. Two new pol sequences were obtained as previously described [26] from infected plasma samples provided by the HIV BioBank. DRM in accordance with the International AIDS Society-USA list 2014 (IAS-USA) [27], were expressed in percentages. Drug susceptibility to 19 antiretrovirals was predicted among those carrying DRM using the Stanford 0 s HIVdb Algorith(http://sierra2.stanford.edu/sierra/servlet/JSierra) in the last available PR sequence from patients carrying DRM, classifying drug susceptibility in four categories depending on mutation scores: susceptible, low-level, intermediate, and high-level resistance.

Statistical analysis
For continuous variables, statistical difference based on the mean was calculated using t-tests. Variable distribution was described as median and interquartile range (IQR). Univariate and multivariate analyses were performed by Weka software (Waikato Environment for Knowledge Analysis, http://www.cs.waikato.ac.nz/ml/weka) to identify the risk factors for DRM development to PI.

LPV/r experienced paediatric population
A total of 199 (37.3%) of the 534 patients from the Madrid Cohort of HIV-infected children and adolescents were exposed to LPV/r within the period 2000-2014. Their epidemiological and clinical features before first LPV/r experience are shown in Table 1 Most were Spaniards (81.9%), diagnosed before 2000 (67.8%), perinataly infected (96.5%) and carrying HIV-1 subtype-B. Last updated data showed that most (85.4%) were under follow-up in paediatric or adult units. The study population showed a mean age 3.2 years at first ART experience, with mono or dual therapy in 55. 3% of cases, a mean age at first LPV/r experience of 9.4 years and a mean time of LPV/ r exposure of 4.2 years.
LPV/r was prescribed as cART in the 199 subjects, although in three (1.5%) of them the ART line was unknown. Among the remaining 196 pediatric patients, LPV/r was included in the first ART regimen in 16.6% patients (group 1), in the second line therapy in 18.6% (group 2), and in third or more line regimen in 63.3% (group 3). The mean time under LPV/r exposure was higher in patients from group 3 (4.8 years) compared with group 2 (3.5 years) or group 1 (3.1 years). The 64 children with available pol sequence or genotypic resistance profile showed a higher mean time under LPV/r experience (5.  Table 1). LPV/r was prescribed more frequently as part of first cART regimen during 2008-2014 than during 2000-2007 (42.8% vs. 11%). The use of LPV/r as off-label antiretroviral was frequent (15.6%) in our study population, as reported for the entire cohort [28]. Table 2 shows the rate of patients with viral suppression according to available VL data from baseline (before LPV/r exposure) in 180 (90.4%) subjects and during LPV/r experience in 194 (97.5%). A total of 3,117 VL measurements after first LPV/r exposure (range from 1 to 48 determinations per patient) were recovered from clinical reports ( Table 3). As expected due to the absence of previous cART exposure, group 1 patients showed higher baseline VL values (5.8 log) than groups 2 and 3 (4.7 log each). Table 3 shows the rate of undetectable VL achieved under LPV/r exposure in the study population. Forty-six (25.5%) of the 180 patients with available baseline VL presented undetectable viraemia (VL< 500c/ml). They belonged to groups 2 (26.1%) and 3 (73.9%). However, nearly 70% of the 194 patients with available viremia data achieved undetectable VL (<500 HIV-1 RNA copies/ml) in the last available VL measurement (Table 3). Considering groups, ART including LPV/r achieved viral suppression (<500 HIV-1 RNA copies/ml) in 8.5%, 68.6% and 64.8% patients from groups 1, 2 and 3, respectively ( Table 2). Group 1 also showed the highest rate of children with undetectable VL (<50c/ml) at last available measurement after LPV/r exposure compared to Group 2 (22/32, 68.5% vs. 17/ 35, 48.57%) ( Table 3).
https://doi.org/10.1371/journal.pone.0173168.t001 DRM to PI and drug susceptibility to LPV/r A total of 64 patients presented resistance information in PR (n = 64) or RT (n = 48) during LPV/exposure (Tables 4 and 5) and 35 of them also at baseline (S1 Table). Among the LPV/r exposed subjects, 27 (42.3%) carried DRM to PI, 28 (58.3%) to NRTI and 21 (43.7%) to NNRTI (Tables 4 and 5). Patients from group 3 presented the highest number of DRM to PI (52.2%), to NRTI (62.5%) and to NNRTI (46.9%). Fig 1 summarizes the DRM in 64 LPV/r experienced paediatric patients in the cohort and in each group and the predicted drug-susceptibility in those 27 patients carrying DRM. Among those 64 patients with resistance data, seven (10.9%) were exposed to LPV/r as offlabel therapy (1 in group 1 and 6 in group 3) and 4 (57.1%) of them carried DRM to PI.
The most common DRM to PI were D30N, M46IL, I54V, V82A and L90M in PR. D30N was the most frequently observed in virus from group 2 individuals and M46I/L, I54V, V82ASTC and L90M in group 3 patients (Fig 1A). Among those 27 viruses with DRM at PR, nelfinavir (NFV) was the most affected PI due to the presence of mutations D30N and L90M. However, darunavir-ritonavir (DRV/r), atazanavir-ritonavir (ATV/r) and tipranavir-ritonavir (TPV/r) presented the highest susceptibility ( Fig 1D, Table 4), due to the absence or low rate of mutations I47V, I50V, I54ML, Q58E, T74P, L76V, V82LT, N83D and I84V in PR ( Fig 1A). As expected, viruses presenting a lower number of DRM to PI at baseline and of previous ART regimens showed a higher predicted susceptibility to PI. However, despite the absence of viruses with DRM in group 1 patients, three patients showed low level resistance to NFV and one child to TPV/r (Table 4). We observed high resistance to fosamprenavir/ritonavir, indinavir/ritonavir, LPV/r, and saquinavir/ritonavir in nearly 50%, to TPV/r in 22.2% and to atazanavir/ritonavir (ATV/r) in 18.5% (Fig 1D). Among those 48 patients with available RT sequence or resistance profile, M184V and T215YF were the most common DRM to NRTI and K103N, Y181C and E138AGK to NNRTI (Fig 1B and 1C). Among those 28 patients with DRM to NRTI the most affected ARVs were abacavir, didanosine, lamivudine and emtrizabine Table 4. DRM presence at PR and predicted drug susceptibility in viruses from LPV/r-exposed patients with resistance data.

Patients with available data at PR a Patients with major DRM at PR; mean DRM [IQR]
Predicted susceptibility to PI (%)   Fig 1D) and among those 21 carrying DRM to NNRTI, efavirenz and NVP, reported the lowest susceptibility (14.3%, both) (Fig 1D). S1 Table shows the DRM to PI profile in the 35 children with available resistance data before and during LPV/r exposure, all with high basal viraemia and LPV/r exposure time.

Risk factors for DRM development to PI
Supervised classification techniques are algorithms with high predictive power and designed to optimize the statistical classification procedures [29]. Supervised techniques using weka analysis identified the use of LPV/r as first ART and year of LPV/r start as the most important clinical factors under study predicting no DRM development to PI in our population (S2 Table). Accordingly, we observed the absence of DRM to PI in viruses from pediatric patients included in group 1.

Discussion
The present study describes the clinical and virological features in patients from the Madrid Cohort of HIV-1-infected-children and adolescents exposed to LPV/r as different-line therapies during 2000-2014. It also reports the VL suppression and lymphocyte recovery after LPV/ r exposure, the presence of DRM to three-drug classes and the predicted drug susceptibility among resistant viruses.
Current guidelines (United States and Europe) support the LPV-based regimen in combination with two NRTI in ARV-naïve children <3 years, and indicate that fosamprenavir, nelfinavir, stavudine, and unboosted atazanavir should not be used for initial therapy [2]. Despite LPV/r being the best choice for starting ART in children of less than three years in Spain, in the study cohort LPV/r was prescribed more frequently within rescue lines than in first-line treatments (83.2% vs. 16.8% cases) during 2000-2014 period. LPV/r was prescribed more frequently as part of the first cART regimen during 2008-2014 than during 2000-2007 (42.8% vs. 11%), probably due to several reasons: i) the lower number of newly infected infants in the first period requiring first therapy according to National guidelines; ii) The alternative efavirenz vs. PI use in children over 3 years with good expected adherence and the preferred ATV/r use due to its single dose use in children over 6 years, which had appeared in a higher number in the first period. iii) The withdrawal of Nelfinavir in 2007 by EMEA.
Although we observed an increase of CD4-T rate in the three groups, patients under first cART including LPV/r (group 1) with the highest number of CD4-T cells counts at baseline, showed significantly higher increases of CD4-T rate in the last available determination during LPV/r exposure, supporting previous studies showing that initiating cART at higher CD4-T cells counts maximizes the immunologic recovery [30]. We observed in the last available VL determination during LPV/r exposure, that 7 of each 10 patients with available data achieved undetectable VL. Viremia reduction was higher (87.5%) in group 1 patients, supporting other studies showing that LPV/r-containing regimens result in potent and durable virological responses in naïve children and in pretreated children [18]. Unfortunately, not all patients in the cohort achieved undetectable VL, since complete suppression of viral replication is more difficult to achieve in children than in adults due to several reasons, including a higher baseline VL, pharmacokinetic issues, the difficulty of permanent adherence because of the bad taste, complex regimens or inappropriate size of pills, and the frequent use of off-label ARV which can lead to under or overdosing [28], increasing the risk of DRM selection. As previously stated, 10.9% of children carrying DRM had been exposed to LPV/r as off-label ART.
Children failing first-line PI-based regimens do so with minimal development of PI resistance [31,32]. In fact, previous reports showed lower DRM appearance after first-line regimens including LPV/r based therapies vs. those including NFV [33] or NNRTI [34]. None of the studied children receiving LPV/r as first ART presented major DRM to PI, which explains the highest VL reduction after LPV/r exposure and preserved susceptibility to most PI observed in this group. As expected, group 3 patients showed the highest DRM rate to the three drug-families, explained by the longest exposure time to drugs. During 2008-2014, EMA approved alternative potent drugs for rescue ART regimens in children, such as new NRTI (tenofovir in 2012), new PI (TPVr and DRV/r in 2009 and ATV/r in 2010) and integrase inhibitor (raltegravir in 2013). According to our data, resistant viruses in LPV/r experienced children showed high susceptibility to TPV/r, ATV/r and DRV/r. Thus, those drugs would be effective after therapeutic failure events in these patients. Previous reports showed that the independent predictive factors related to virological success of LPV/r based ART regimens were plasma viraemia levels, previous PI use and the number of mutations reducing susceptibility to LPV/r at baseline [35,36].
In our study, 11 of the 37 children with no DRM to PI and resistance data before and after LPV/r exposure did not develop DRM at PR after a mean time of LPV/r exposure of 4.3 years (S1 Table). However, since none of them presented DRM to RT inhibitors, the high VL observed in the last available determination in three patients with known viremia would suggest bad adherence to treatment. The high rate of DRM to all drug classes among the 64 patients with available resistance data during LPV/r treatment is in agreement with the long history of therapy, frequent regimen switches and drug experience, and high off-labels ARV exposure, as previously reported [24,28,37,38]. Some limitations to the study are the relatively late initiation of LPV/r in this cohort (9.5 years), the high rate of patients including LPV/r as at least third line treatment (with severe failure background), the high evolution of ARV options in children and adolescents during the study period and the late formal approval of LPV/r use in children. Moreover, only 64 of the 199 patients with LPV/r exposure during 2000-2014 presented available pol sequences or resistance profiles, favoured by the non recommendation of resistance testing in naïve subjects during clinical routine by the Spanish guidelines until 2007 [39] and to the low availability of samples before or after LPV/exposure. This study describes the clinical follow-up in a cohort of LPV/r experienced children and adolescents during 15 years in Spain. A higher VL reduction and better CD4 and CD8 recovery was observed when LPV/r was taken as part of a first-line regimen, reinforcing its use in children under 3 years old. LPV/r can also be a good alternative for rescue regimens in HIV-infected children with previous failure to NNRTI, and when other new PI are not available or present inadequate paediatric formulations for children less than 6 years. However, since most paediatric patients requiring a third ART line have been under PI exposure, LPV/r would not be a good candidate in this scenario.
Supporting information S1 Table. DRM in 35 HIV-1 infected children with available data before and during LPV/r exposure. Patients without asterisk refers to the last baseline PR sequence before LPV/r exposure and marked with asterisk the last available PR sequence or resistance profile collected during LPV/r treatment until December 2011. Date: day/month/year. DRM, drug resistance mutations according to IAS-USA 2014 using Stanford 0 s HIVdb Algorithm (http://sierra2. stanford.edu/sierra/servlet/JSierra); PI, protease inhibitors; NA, data not available. Among these 35 patients, 11 carried wild type viruses in the analyzed baseline and post-LPV/r exposure sequences after a mean time of LPV/r exposure of 4.3 years. In 7 subjects infected by wild type viruses, major DRM to PI appeared after a mean time of LPV/r use of 7.3 years and a mean interval between sequences of 5.7 years. Seventeen patients were infected with viruses carrying major DRM to PI at baseline, maintaining resistant viruses at PR during their entire follow up in 10 cases or reverting to wild type viruses in 7 cases after a mean time of LPV/r use of 5.6 years. Five patients maintained the same DRM-PI profile in both sequences collected in a mean interval of 1.5 years and after a mean time of LPV/r exposure of 3.7 years, all with detectable and high VL at sampling time. (DOCX) S2