A randomized controlled trial comparing peginterferon-α-2a versus observation after stopping tyrosine kinase inhibitor in chronic myeloid leukemia patients with deep molecular response for at least two years

Background: Interferon (IFN) is a logical possibility to increase treatment free remission (TFR) rate in chronic myeloid leukemia (CML). We conducted the rst randomized controlled trial comparing the use of pegIFN versus observation in CML patients attempting TFR. Methods: Adult CML patients with stable deep molecular response for ≥ 2 years with ≥ 2 readings of MR4.5 were randomized into two arms -- subcutaneous pegIFN-α-2a starting at 180µg weekly for a year, followed by observation, or observation. Results: A total of 30 patients were recruited (pegIFN n=15, observation n=15). Median follow-up was 38.1 months (range 15.9-49.4) and 23.8 (1.5-51.0) in pegIFN and observation arm, respectively. A total of 11 patients relapsed (pegIFN n=4, observation n=7). The median time of relapse was 13.1 months (range 9.2 to 25.5) and 4.4 (1.2 to 13.6) in pegIFN and observation arm, respectively. Only 8 out of 15 (53%) patients completed 52 doses of pegIFN with mean dose of 43 out of 52 doses (range 20 to 52). Dose of tolerable pegIFN was age dependent. Conclusion: PegIFN is a potential consolidative therapy to increase TFR.

Chronic myeloid leukemia (CML) is a myeloproliferative neoplasm that originated from an abnormal pluripotent bone marrow stem cell and is consistently associated with BCR-ABL1 fusion gene. (1) The level of BCR-ABL1 fusion gene can be quantitated using real-time quantitative polymerase chain reaction (qPCR) and is standardized using International Scale (IS) (qPCR IS ).(2) BCR-ABL1 fusion gene is translated into an abnormal tyrosine kinase, a key player in pathogenesis of CML.
The rst tyrosine kinase inhibitor (TKI) is imatinib (Glivec® or Gleevec®, Novartis Pharmaceuticals Corporation). TKI changes the paradigm of CML from a doomed incurable disease unless allogenic hematopoietic stem cell transplantation was done, to a disease that most patients can have a near normal life span. It also revolutionizes the approach in oncology from chemotherapy/radiotherapy to targeted therapy. Since imatinib was approved by United State Food and Drug Administration in 2001 (3), there were many advancement in the management of CML. One of them is the concept of treatment free remission (TFR), arguing the necessity of life long TKI. Stopping TKI treatment was reported previously(4-12), but it was not landscape changing until the landmark stop TKI trials from French (13) and Australia (14) reported about 40% of CML patients, who had achieved complete molecular response (CMR), i.e. undetectable BCR-ABL1 gene from qPCR, for at least two years, were able to stop TKI safely and remain in TFR.
Interferon (IFN)-α, either IFN-α-2a or IFN-α-2b, is the standard treatment of CML before the era of TKI. Depending on the duration of administration, only about 10 to 30% of patients on IFN-α(15) achieved complete cytogenetic response compared to about 70 to 90% on imatinib (16). However, IFN acts differently from TKI. It might be able to target leukemic stem cells. (17) Case reports and case series show that IFN-responded patients could achieve long term TFR. (18)(19)(20)(21)(22)(23)(24) In the largest case series, 78 to 95% of IFN-responded CMR patients remained in TFR once IFN was withdrawn (25,26), probably via IFN-induced immunity towards the leukemic stem cell clone. The role of IFN in TFR is further supported by studies incorporating IFN or its pegylated form (pegIFN) as upfront treatment together with TKI (27), upfront treatment together with TKI and consolidation (28,29) and consolidation after stopping TKI (30)(31)(32)(33). These studies showed higher TFR rate compared to the landmark stop TKI trials (13,14). However, de nitive conclusion that IFN/pegIFN can increase TFR could not be drawn from these studies because they were all single arm study.
We conducted a randomized controlled trial to compare the outcomes of pegIFN-α-2a for a year followed by observation versus observation after stopping TKI in CML patients with deep molecular response (DMR) (molecular response (MR) of 4-log reduction (MR4) (0.01% IS ) or deeper) for two years or more. To our knowledge, this is the rst randomized controlled trial to study the effect of IFN-α to TFR. We reported the result of this pilot study here.

Study design and patient eligibility
This was a multi-center randomized open-label controlled trial conducted in Malaysia. This study was approved by Medical Research and Ethics Committee, Ministry of Health Malaysia. Written informed consent was obtained for each participant prior to the enrollment.
Inclusion criteria were 1) 18 years old or above, 2) chronic phase during diagnosis of CML, 3) treated with ongoing TKI at any dose for at least 3 years, 4) achieved stable DMR for two years or more by any TKI, and 5) the date of the latest qPCR IS result must compel with Intervention Start Date (ISD). Stable DMR was de ned as DMR with at least two results, including the latest, of MR4.5 (0.0032% IS )(2) or deeper over the last two years. Study Entry Date was de ned as the date patient signed consent form. ISD was de ned as the day after the last dose of TKI and is within four weeks of Study Entry Date or 17 weeks from the latest qPCR IS test. Exclusion criteria were elaborated in Supplementary Data. Randomization process, allocation concealment and pegIFN-α-2a are elaborated in Supplementary Data. Patients were randomized into two arms, one was subcutaneous pegIFN-α-2a, starting at 180 µg weekly, for one year followed by observation and one was observation throughout. The next day after the last dose of TKI, patients would receive either one of the two arms according to the randomization result.

Study outcomes
The primary outcome was relapse rate and relapse-free survival (RFS). Relapse was de ned as: 1) one reading of loss of major molecular response (MMR) (0.1% IS ), or 2) positivity of BCR-ABL1 transcripts in qPCR IS , as con rmed by a second analysis point, indicating the increase (≥ 1 log) in relation to the rst analysis point at two successive assessments. Time of Relapse (ToR) is de ned as the time of loss of MMR (relapse criteria no.1) or of the rst analysis point (relapse criteria no.2), whichever is earlier. After relapse, the treating physician was recommended to restart the patient on TKI and required to monitor patient monthly until regain DMR (DMR 2 ) consecutively for two months, following which patient will return to the regular monitoring every 3 to 6 months.
The secondary outcomes were 1) Time to Relapse (TTR) (time from ISD to ToR), 2) DMR 2 rate, 3) Time to DMR 2 (ToR to time of rst analysis point of DMR 2 ), 4) adverse side-effects of peginterferon and 5) quality of life (QoL) assessment.

Data collection, monitoring and safety precautions
Peripheral blood for qPCR IS for BCR-ABL1 test (see Supplementary Data) was centralized and performed following the standard operating procedures in Molecular Laboratory, Ampang Hospital. Patients came for monthly qPCR IS monitoring for the rst 12 months, 2-monthly for subsequent 12 months, and 3monthly thereafter (see Supplementary Data for other investigations and QoL assessment). Repeating the tests before the scheduled time interval was not allowed unless it is deemed indicated after discussed with the Principle Investigator.

Statistical analysis
Comparison of characteristics of the patients were performed using Chi-square test for categorical data and ANOVA or Kruskal Wallis test for normally and non-normally distributed continuous data, respectively. RFS was analyzed using log rank test. Rate were analyzed using Chi-square test (Fisher's Exact Test if the count in the cell is less than expected) and risk estimation. Comparison of duration was analyzed using independent T test. Descriptive analysis was used to analysis number of patients who developed side effect. Parametric tests for normally distributed data and non-parametric tests for skewed distribution were used to analyze QoL. Signi cant level was set at 0.05. Statistical Package for Social Science (SPSS) 24 software was used for statistical analysis of this study.

Page 6/20
The recruitment was started in March 2015 and closed in Sep 2018. ISD of the rst and last recruited patient was 8th July 2015 and 7th October 2018, respectively. Analysis was done as per status on 7th October 2019. Data from a total of 30 patients was analyzed. Median follow-up was 33.0 months (range 1.5-51.0). The overall study ow diagram is shown in Fig. 1.
All patients were on imatinib before starting intervention, except one patient (pegIFN arm) was initially on imatinib but later changed to nilotinib due to imatinib-induced diarrhea. The baseline characteristics of the 30 patients is shown in Table 1. There is a difference in gender distribution between the two arms.  . 2). In comparison to relapse patients in observation arm, relapse patients in pegIFN arm also had shorter duration of TKI (p = 0. 28) and shorter duration of MR3 (0.1% IS ) or deeper (see Fig. 2).
In general, QoL (see Table 2) in both arms were similar at baseline and after two months stopping TKI.
Symptoms related to TKI side effects like nausea and vomiting and diarrhea were reduced after stopping TKI in both arms. Safety and side effects Among the 23 reasons of pegIFN dose reduction/withholding, 12 (52%) were due to liver transaminitis, followed by blood count changes 4 (17%), intolerance like giddiness and pruritus 3 (13%), symptoms of carpal tunnel syndrome 2 (9%), raised blood pressure 1 (4%) and thyroid dysfunction 1 (4%). Among the 15 patients on pegIFN arm, the highest alanine transaminase (ALT) rise was 7 times upper limit of normal (ULN) (n = 1), followed by 6 times (n = 2), 5 times (n = 1), 2 times (n = 6), 1 time (n = 2) and never raise more than 1 time (n = 3). Five patients needed to withhold pegIFN due to ALT ≥ 5 times ULN (n = 4) and platelet < 20 × 10 9 /L (n = 1). The ALT rise of ≥ 5 times ULN in three out the four patients could have been avoided if the starting dose of pegIFN followed the age of the patients (see Discussion). All the transaminitis resolved after dose reduction/withholding.
TKI withdrawal syndrome (reported myalgia and arthralgia) occurred more in observation arm 8 (53%) compared to pegIFN arm 5 (33%), which is also re ected in QoL (see Table 2). The reported myalgia and arthralgia were grade 1 except one patient had grade 2 and one patient was severe requiring withdrawal from the study (see Supplementary Data). One death but was unrelated to CML or the study (see Supplementary Data).

General limitation of the study
The main limitation is the small sample size. However, if a signi cant nding was found in a small sample, like our pilot study, it encourages further and larger study. The lower relapse rate 4 (27%) versus 7 (47%) and later TTR 12.9 versus 1.8 months in pegIFN arm versus observation arm, respectively, might provoke argument that the nding merely because, as a form of CML treatment, pegIFN delays the relapse. Thus, the duration of follow-up must be long enough to overcome this potential confounder. We reported our study after the last recruited patient completed a year of follow-up and by chance, the pegIFN arm had longer follow-up 38 From this study, we found the likely tolerable pegIFN-α-2a dose is correlated with age (see Supplementary  Fig. 1). Among 13 patients who required pegIFN dose reduction, 9 (69%) patients could have avoided/reduced the need of dose reduction if the starting dose was lower. This would certainly help future similar study to achieve the target accumulative dose of pegIFN-α-2a, reduce unnecessary side effects, and improve compliance. We would suggest the starting pegIFN-α-2a dose as 180 µg weekly in patients age below 40 years old, 135 µg weekly in patients age 40 to 60 years old and 90 µg weekly in patients age more than 60 years old in future similar study.

Fluctuation of qPCR IS level
Fluctuation of qPCR level below MMR (0.1% IS ) is a known phenomenon after stopping TKI.(40) To our knowledge, there was no report on uctuation that exceeding MMR, which is probably the reason it is recommended as a criterion of relapse (41) and used in most of the stop TKI trials (42). We would like to report two cases of uctuation exceeding MMR here.
Two patients in the observation arm, both from the same study site, relapsed at different time point (one at 12 months (Dec 2017) and one at 14 months (Feb 2018) after stopping TKI) according to the relapse criteria no.1, i.e. loss of MMR. TKI was reinitiated as per protocol. However, a repeated qPCR IS , which was not prohibited in the study protocol, was done prior to the initiation of TKI, which showed DMR.
Investigations showed no evidence of wrong sampling or laboratory error. After discussion, investigators decided to stop their TKI after two months of TKI intake. These two patients remained in TFR up to the time of writing. Their qPCR IS trend and blood count changes compared to the other two relapsed patients at the same study site and other study sites are shown in Supplementary Fig. 2 and Supplementary Fig. 3, respectively.
These two "relapse" cases challenge MMR as a relapse criterion and raise doubt on the four relapse cases prior to the incidence. We re-examined these four cases and could only truly con rm relapse in one case (observation arm), in which the previous two readings already showed 1-log increment from 0.0059% IS to 0.0896% IS , ful lled relapse criterion no.2, before loss MMR with reading of 0.2594% IS . We could not rule out the possibility of uctuation in other three cases con dently because no repeat qPCR IS was done before restarting TKI. One patient (observation arm) had two prior readings of 0.0117% IS and 0.0774% IS before loss of MMR 0.1931% IS . One patient (pegIFN arm) had two prior readings of 0.0225% IS and 0.0680% IS before loss of MMR 0.1085% IS . One patient (pegIFN arm) had two prior readings of 0.0203% IS and 0.0664% IS prior to loss of MMR 0.1653% IS . Following the incidence, study protocol was amended to include a repeated qPCR IS on the time of restarting TKI after loss of MMR, which means relapse criteria no. 1 -loss of MMR -must be con rmed by two successive readings, and re-stopping TKI if the repeated qPCR IS does not con rm the loss of MMR. In fact, at the time of writing (Apr 2020), there was another patient (pegIFN arm), who had such uctuation, with two prior reading of 0.0108% IS and 0.0337% IS before loss of MMR 0.1046% IS and the repeated reading of 0.0232% IS . He was restarted and planned on TKI for two months, just like the previous two cases of uctuation, but had not came back for review due to Malaysia and Singapore lock-down during COVID-19.
Gender effect on the e cacy of pegIFN?
The limitation of this pilot study is small sample size, which caused the signi cant difference in the gender ratio between the two arms. PegIFN arm had more male patients whereas observation arm had more female patients. Earlier single arm observation study (13) suggested female is less likely to maintain TFR, but the nding was disputed after a longer follow-up (43). Other single arm observation studies also did not nd gender as a predictor of TFR. (39,41,(44)(45)(46) Thus, it is unlikely gender distribution affects the outcome in the observation arm, but we cannot extrapolate the same assumption to pegIFN arm. In fact, there are data to suggest pegIFN is less effective in female compared to male in the treatment of Hepatitis C using pegIFN and ribavirin.(47-49) A signi cant more female in relapse pegIFN arm compared to non-relapse pegIFN arm (p = 0.03) in our small sample size warrants careful interpretation of the outcomes.

TKI and DMR duration
Relapse patients in pegIFN arm had shorter duration of TKI and shorter duration of DMR compared to non-relapse patients in pegIFN arm and either relapse or non-relapse patients in observation arm (see Fig. 2). The clinical predictors of TFR found in many single arm observation studies are longer duration of TKI and/or longer duration of DMR. (39,43,45,46) This factor should be considered during interpretation of the outcomes.

Other ndings
First, the increment of the blood counts during the rst two visits and plateau off after that period in the observation arm (see Supplementary Fig. 4) provides further evidence for the myelosuppression effect of TKI. It also explains partly the improvement in QoL (see Table 2) and justi es the need of TFR. Second, there was a groove in the trend of the blood counts in the pegIFN arm during the six months stopping TKI corresponding to the adjustment of pegIFN dosage during that period (see Supplementary Fig. 4). However, the blood count trend and the use of pegIFN did not affect the QoL (see Table 2). Third, absolute basophil count (ABC), but not total white cell count (TWC), was shown to correlate with percentage of Philadelphia chromosome in marrow metaphases of pre-clinical CML.(50) However, it seems that ABC also correlate with qPCR IS (see Supplementary Fig. 3 (c)) at lower disease burden as in pre-clinical CML.
TWC seems to correlate with qPCR IS , too. These ndings were unexpected. Thus, we examined whether these correlations could be used as surrogate marker of the uctuation. The data is immature, but TWC and ABC might be the surrogate marker of the uctuation (see Supplementary Fig. 3 (a)).

Conclusion
In conclusion, this is the rst randomized controlled trial studying whether pegIFN could increase TFR rate compared to observation. PegIFN is a potential consolidative therapy to increase TFR.  Figure 1