Author Correction: The Impact of Liposomal Irinotecan on the Treatment of Advanced Pancreatic Adenocarcinoma: Real-World Experience in a Taiwanese Cohort

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Oral S-1, a third generation oral fluoropyrimidine, is another useful and approved agent for the treatment of pancreatic cancer in Asia. S-1 monotherapy has been shown to achieve significantly better response rate and non-inferior overall survival against gemcitabine alone for advanced PDAC patients in the randomized phase III GEST trial 12 . For its favorable safety profile, S-1 either alone or combining with gemcitabine are acceptable regimens for less fit, advanced PDAC patients in Japan 10,13 and for all comers in clinical practice before the reimbursement of FOLFIRINOX and the nab-paclitaxel in Taiwan. While biweekly triplet chemotherapy consisting of gemcitabine, oxaliplatin plus either infusion 5-FU/LV (the GOFL regimen) or oral S-1/LV (the SLOG regimen) that have been developed through a series of multicenter trials under the platform of Taiwan Cooperative Oncology Group (TCOG) are the favorable regimens in our institutional clinical practice 14,15 . In a single arm phase II trial, SLOG had 40.7% of overall response rate, and 7.6 and 11.4 months of median progression-free survival and overall survival, respectively, that are comparable with those achievable with FOLFIRINOX in ACCORD 11/0402 study 15 .
There was no consensus regarding second-line treatment for pancreatic cancer until 2015, in which global NAPOLI-1 trial showed liposomal irinotecan plus 5-FU/leucovorin (nal-IRI + 5-FU/LV) not only significantly extended survival but also preserved quality of life in patients with gemcitabine-refractory metastatic PDAC as compared to 5-FU/LV 16,17 . Nal-IRI is a new formulation of irinotecan encapsulated in pegylated liposomes which prevents the premature metabolism of irinotecan in the liver and can passively diffuse through leaky tumor vasculature, resulting in favorable circulation and intra-tumor pharmacokinetics of SN-38, the active metabolite of irinotecan, as compared to conventional irinotecan 18,19 .
Aside from the pivotal NAPOLI-1 trial, real-world data reporting the efficacy and safety of nal-IRI + 5-FU/ LV in gemcitabine-refractory advanced PDAC are scarce [20][21][22] . Herein, we report the experience from National Cheng Kung University Hospital (NCKUH), a high-volume medical center for pancreatic cancer management in southern Taiwan 23 .

Methods
Patients. This retrospective study included 44 consecutive gemcitabine-refractory, advanced PDAC patients who had nal-IRI + 5-FU/LV treatment at NCKUH between December 2016 and December 2018. The median follow-up time, from the date of starting treatment to the date of cutoff at July 1st, 2019 was 19.7 months. Patient demographics, clinicopathological characteristics, dates of confirmed PDCA diagnosis and detecting metastatic disease, all previous treatments before nal-IRI + 5-FU/LV, date of starting nal-IRI + 5-FU/LV, dates of disease progression after nal-IRI + 5-FU/LV and death or last follow-up as well as all treatment-related adverse events were recorded through electronic medical records review. Generally, tumor assessment was done by computed tomography (CT) before treatment and every 12 weeks after treatment mainly based on national health insurance regulation and the in-charge physician's discretion. Tumor response was assessed by RECIST version 1.1 and patients who presented obvious clinical disease progression were classified as having progressive disease (PD) if no follow-up images were available. Statistical analysis. Descriptive statistics were applied to demographics and presented as median or percentage, as appropriate. Progression-free survival (PFS) was calculated from the date of starting nal-IRI + 5-FU/ LV treatment to the date of disease progression or death/loss follow-up, whichever occurred first. The time of PD was defined as the date of radiographic progression as judged by a radiologist or clinical progression warranting a change in or termination of treatment, as judged by the treating physician. Survival was determined by the Kaplan-Meier method, and differences in survival between groups were compared using the log-rank test. All variables with p < 0.05 were considered to be statistically significant. All statistical analyses were performed using R version 3.5.1. This retrospective analysis was approved by the National Cheng Kung University Hospital (NCKUH) Institutional Review Board (No. A-ER-108-113) with waiver of informed consent and followed the Helsinki Declaration.

Patient characteristics.
A total of 44 patients with advanced PDAC who received at least one dose of nal-IRI + 5-FU/LV at NCKUH between December 2016 and December 2018 were identified. The baseline demographics, clinicopathological characteristics and dose delivery are listed in Table 1. Their median age was 60 years old (range 26-80) and 25 patients were male (56.8%). Baseline characteristics before starting nal-IRI + 5-FU/ LV were Eastern Cooperative Oncology Group performance score (ECOG PS) 0-1 in 32 (72.7%), presence of metastatic diseases in 39 (88.6%) with 56.8% exhibiting liver metastasis and failed to one prior chemotherapy in 32 (72.7%). The most common prior regimen was gemcitabine-based triplet therapy consisting of gemcitabine, oxaliplatin plus either 5-FU/LV or S-1/LV 14,15 in 24 (54.5%); while only 4 patients (9.1%) had prior exposure to irinotecan.
Drug delivery. Besides the inclusion of 12 (27.3%) patients of ECOG PS ≥ 2, the other unique feature of this study was 29 (65.9%) patients had lower initial dose of nal-IRI. The starting dose of nal-IRI was reduced by 20% to 60 mg/m 2 (equivalent to 52 mg/m 2 irinotecan free-base) in 19 of 32 patients with ECOG PS 0-1 and 10 of 12 patients with ECOG PS ≥ 2 aiming to improve the treatment compliance at the discretion of in-charged physicians. The median baseline albumin level of patients with the standard 80 mg/m 2 (equivalent to 70 mg/m 2 irinotecan free-base) and reduced 60 mg/m 2 starting dose groups were 4.1 and 3.4 g/dl, respectively (Table 1). A summary of nal-IRI delivery is listed in Table 2. The median treatment duration and administration cycle in the standard and reduced starting dose groups was 14.6 weeks and 6 cycles, and 9.6 weeks and 4 cycles, respectively. The average interval between treatment cycles was 2.9 and 3.0 weeks in corresponding groups. Six (20.6%) patients with 60 mg/m 2 starting dose had dose escalation to 70-80 mg/m 2 during their treatment course. The www.nature.com/scientificreports www.nature.com/scientificreports/ median 6-week cumulative dose in ECOG PS 0-1 patients of standard and reduced starting dose and ECOG PS 2 were 216, 192 and 116 mg/m 2 , respectively ( Fig. 1). Table 3. The median PFS was 2.5 months (95% confidence interval [95% CI]: 2.2-5.8 months) and median OS was 6.6 months (95% CI: 3.8-9.5 months) for the entire cohort. The overall response rate of entire cohort was 9.1%. Two of four responders, locally advanced and metastatic PDAC in one each, who failed to SLOG had successful conversion surgery after second-line nal-IRI + 5-FU/LV. Both patients achieved more than one year of relapse-free survival after surgery.

Treatment response. Therapeutic efficacy is summarized in
A total of 31 patients had evaluable target lesions; the percentage of change in tumor size is depicted in Fig. 2. Of the entire cohort, the median OS was 8.4 months in patients with partial response (PR) or stable disease (SD), and 3.2 months in the PD group (Fig. 3A). The median OS was 7.8 months for the 32 ECOG PS 0-1 patients and 2.7 moths for the 12 ECOG PS ≥ 2 patients. The median OS of the ECOG PS 0-1 patients with and without initial dose reduction were 8.4 and 7.5 months, respectively (p = 0.95), as shown in Fig. 3B. www.nature.com/scientificreports www.nature.com/scientificreports/ The duration of treatments before and after nal-IRI + 5-FU/LV was available in 42 patients. Twenty-four out of 42 patients (57.1%) with advanced PDAC survived for more than one year, calculated from the start of first-line treatment (Fig. 4). The median OS after the first radiological evidence of metastatic diseases was 13. Safety profiles. Safety profiles that were extracted from electronic medical records included patient symptoms, physical examination, complete blood count and biochemistry lab data ( Median (range) 14.6 (0-58.0) 9.6 (0-80.6) 9.9 (0-80.6)

Discussion
Current study shows that nal-IRI + 5-FU/LV regimen could achieve a 9.1% of response rate and 6.6 months of overall survival for gemcitabine-refractory advanced PDAC on intent-to-treat analysis in a real-world Taiwanese cohort, including 27.3% ECOG PS > 2 patients and 65.9% having reduced initial dose of nal-IRI to 60 mg/m 2 . The treatment outcomes largely concur with those of pivotal NAPOLI-1 trial in which most patients were ECOG PS 0-1 and had standard starting dose of 80 mg/m 2 nal-IRI 16 . After excluding those with ECOG PS > 2, the median OS of our ECOG PS 0-1 cohort was 7.8 months, which was compatible with that of 8.9 months in Asian  Table 3. Treatment efficacy. *Includes clinical progressive disease lacking CT scan evaluation.  www.nature.com/scientificreports www.nature.com/scientificreports/ subgroup of NAPOLI-1 24 . To the best of our knowledge, the clinical outcomes of ECOG PS > 2 patients receiving nal-IRI + 5-FU/LV remains yet illustrated, in neither previous reports from the Memorial Sloan Kettering Cancer Center (MSKCC) and NAPOLI-1 studies that including 21% and 9% of ECOG PS 2 patients, respectively 16,20 . In this single-institution analysis, the median OS of 2.7 months for ECOG PS ≥ 2 group suggests that although nal-IRI + 5-FU/LV could be well tolerated, but its routine use in patients of gemcitabine-refractory advanced PDAC with ECOG PS ≥ 2 should not be encouraged.
In the pre-specified extended analysis of NAPOLI-1 study, only 41.2% of Asian patients with nal-IRI + 5-FU/ LV could have 80% or more of scheduled dose intensity during the first 6 weeks of treatment as 65.3% non-Asian patients did 25 . In addition, a post hoc analysis of NAPOLI-1 showed Asian patients with nal-IRI + 5-FU/LV had a 54.5% of grade 3-4 neutropenia as compared to the 17.8% in Caucasian patients 24,26 . Considering the higher incidence of hematological toxicities and thus more dose modification in the nal-IRI + 5-FU/LV treated Asian population of NAPOLI-1 study [24][25][26][27] , our physicians chose to give lower starting doses of nal-IRI at 60 mg/m 2 in some patients with poor physical performance and/or poor nutrition status. This is reflected by the fact that patients in the reduced dose group were more of ECOG PS > 2, 34.5% versus 13.3%, and had lower baseline median albumin level, 3.4 g/dl versus 4.1 g/dl, as compared to those in the standard dose group. The strategy seems to work because among those with 60 mg/m 2 nal-IRI starting dose, only 13.8% patients required further dose reduction, while  www.nature.com/scientificreports www.nature.com/scientificreports/ 20.6% patients could have dose escalation to 70-80 mg/m 2 in their subsequent treatment cycle(s). Although mean dose intensity of 57.7% in the 60 mg/m 2 starting dose group was lower than that of 68.8% in standard starting dose group, however, the survival of ECOG PS 0-1 patients with and without starting dose reduction was compatible (Fig. 3B). The impact of early dose modification has also been evaluated in a post hoc analysis of NAPOLI-1, in which all non-UGT1A1*6 homozygous patients received standard starting dose of nal-IRI. Among the 93 patients who had >6 weeks of scheduled nal-IRI + 5-FU/LV treatment, the median OS of those with and without early dose modification (reduction and/or delay within the first 6 weeks of treatment) was 8.4 and 8.3 months, respectively 28 . These data highlights that early dose modification of nal-IRI may not significantly affect the clinical outcomes of patients having nal-IRI + 5-FU/LV for gemcitabine-refractory advanced PDAC, regardless of the modification was made at or after the start of nal-IRI + 5-FU/LV treatment.
Since nal-IRI is a new formulation of irinotecan, it is conceived that patients with history of prior irinotecan exposure might be less likely to benefit from nal-IRI-based treatment. In both the NAPOLI-1 and a MSKCC studies, irinotecan-naïve patients were benefited more from nal-IRI + 5-FU/LV treatment as compared to those with prior irinotecan exposure with HR of 0.62 (95% CI: 0.44-0.86) and 0.38; 95% CI: 0.2-0.72), respectively 16,20 . These findings raise the concern regarding the clinical relevant of nal-IRI + 5-FU/LV in daily practice when FOLFIRINOX appears to be one of recommended first-line regimens for chemo-naïve, ECOG PS 0-1 advanced PDAC patients in all recent NCCN, ASCO and ESMO pancreatic cancer management guidelines [7][8][9]11 . However, due to its association with high hematological toxicity, first-line FOLFIRINOX could only be applied in 20-40% of advanced PDAC patients with good performance, as demonstrated in several recent real world population analyses [29][30][31] . On the other hand, gemcitabine-based regimens remain the treatment of choice for the rest of 50-60% chemo-naïve advanced PDAC patients. Among them, the use of gemcitabine plus nab-paclitaxel reached 40% after 2014 in an US population study, 47% in a Pan-European Questionnaire Study and 56.8% in a Japanese study that evaluated the dynamical changes of treatment patterns in real-world practice [29][30][31] . Nal-IRI + 5-FU/LV remains the only FDA and EMEA approved regimen for advanced PDAC patients who have failed to gemcitabine-based therapy.
In the past, surgery did not improve the survival of those highly selected patients with solitary liver metastasis and thus rarely considered as an treatment option for patients with metastatic PDAC 32 . With the advance of front-line multi-agent combination regimens, there is an increasing interest in conversion surgery for patients with locally advanced PDAC and very occasionally for patients with metastatic PDAC 33,34 . In current study, two of four patients with partial response after nal-IRI + 5-FU/LV treatment was able to undergo successful conversion surgery followed by maintenance nal-IRI + 5-FU/LV. Both achieved more than one year of relapse-free survival after surgery. Such experience suggests the feasibility and potential benefit of conversion surgery in advanced/ metastatic PDAC patients even after second-line treatment.
Of patients in the nal-IRI + 5-FU/LV arm of the pivotal NAPOLI-1 study, Asian cohort experienced higher incidence of grade 3-4 neutropenia but lower incidence of grade 3-4 diarrhea as compared to the overall study population 16,24 . Primary granulocyte colony-stimulating factor prophylaxis was not routinely administrated in our institute. The 27.4% of grade 3-4 neutropenia in our patients was lower than that of 54.5% in the NAPOLI-1 Asian cohort, which can be partially attributed to the use of reduced starting dose of nal-IRI in 65.9% of our patients as well as lower mean dose intensity of nal-IRI in our patients, 57.7-68.6% versus 74.9% in the NAPOLI-1 Asian cohort 24 . None of our patients experienced grade 3-4 diarrhea further supports the observation that nal-IRI-based therapy-related grade 3-4 diarrhea less frequently occurs in Asian population as compared to non-Asian and its incidence is nal-IRI dose dependent regardless of ethnicity in earlier studies 26,27 . For instance, despite same dose intensity of nal-IRI in nal-IRI monotherapy (120 mg/m 2 , every 3 weeks) and nal-IRI + 5-FU/ LV (80 mg/m 2 , every 2 weeks), the incidence of grade 3-4 diarrhea was higher in nal-IRI monotherapy, 21.1% versus 12.8% in overall population and 16.0% versus 3.0% in Asian population of NAPOLI-1 study 24 . A population pharmacokinetic analysis of earlier nal-IRI studies found the ethnic difference of grade 3-4 neutropenia and diarrhea can be associated with the Cmax level of uncapsulated SN38 and total irinotecan, respectively 26 . Unfortunately, the genetic background under such pharmacokinetic differences remain largely unknown and deserve further investigation.
There are some limitations in our retrospective study that are also common in other real-world data analysis. First, the inclusion of fragile patients of ECOG PS > 2; secondly, more labile treatment and dose modification schedule in daily practice; and thirdly, data extraction from electronic medical records which may not comprehensive and accurate as those from clinical trials. However, our study not only reflects the experience of real-world practice in a high-volume medical center in Taiwan, but also provides evidence for the feasibility and efficacy of nal-IRI + 5-FU/LV for ECOG PS > 2 patients.

conclusion
Our study confirmed the therapeutic efficacy of nal-IRI + 5-FU/LV treatment with acceptable toxicity profiles in a Taiwanese cohort. However, the benefits for patients with ECOG PS ≥ 2 remained unremarkable. Our experience supports the claim that nal-IRI + 5-FU/LV could serve as standard care after the failure of first-line gemcitabine-based treatment in patients with advanced PDAC in real-world clinical settings.