Is there an Exposure–Response Relationship for Nivolumab in Real-World NSCLC Patients?

Round 1

Reviewer 1 Report

The authors have revised the manuscript according to my suggestions. In my opinion, the manuscript is  for publication in Cancers.

Reviewer 2 Report

I have no further comments. 

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

The meaning of "PK/PD relationship" is little bit difficult to understand. However, the present results are clear. 

Author Response

We are grateful for your valuable contribution to improve our manuscript.

The meaning of "PK/PD relationship" is little bit difficult to understand. However, the present results are clear. 

We agree with you. We changed the title. The new one is “Is There Exposure-Response Relationship for Nivolumab in Real World NSCLC Patients?”

Reviewer 2 Report

I have read your manuscript with interest and reviewed the literature related to it. The investigation of the PK/PD data from a large cohort of NSCLC patients to unravel new predictive/prognostic factors is an interesting topic that warrants further investigation. The article shows an impressive data collection of 68 NSCLC patients with a clear experimental set-up, and convincing data on the deleterious effect of corticosteroids for nivolumab treatment. I have a few suggestions to improve the quality of the manuscript:

A significant correlation between Cmin and use of corticosteroids was seen at day 28, but not at day 42. Do the authors have an explanation for this? This should be included in the discussion

PDL1 testing data were available for 43 patients, however no associations were made between Cmin and PDL1 expression in the tumor. If possible, this should be added to the table.

No subgroups analysis were assessed. For example Cmin - Stage IV vs Stage III. Cmin - Adenocarcinoma/Squamous. Cmin - Cerebral metastasis yes/no. This should be included.

218: 'simulation has shown'

Author Response

We are grateful for your valuable contribution to improve our manuscript.

I have read your manuscript with interest and reviewed the literature related to it. The investigation of the PK/PD data from a large cohort of NSCLC patients to unravel new predictive/prognostic factors is an interesting topic that warrants further investigation. The article shows an impressive data collection of 68 NSCLC patients with a clear experimental set-up, and convincing data on the deleterious effect of corticosteroids for nivolumab treatment. I have a few suggestions to improve the quality of the manuscript:

A significant correlation between Cmin and use of corticosteroids was seen at day 28, but not at day 42. Do the authors have an explanation for this? This should be included in the discussion

The lack of significant association between nivolumab plasma Cmin at D42 and corticosteroids intake could be related to a lower statistical power because of many missing data. We added this sentence in the revised discussion (lines 270-273).

PDL1 testing data were available for 43 patients, however no associations were made between Cmin and PDL1 expression in the tumor. If possible, this should be added to the table.

No subgroups analysis were assessed. For example Cmin - Stage IV vs Stage III. Cmin - Adenocarcinoma/Squamous. Cmin - Cerebral metastasis yes/no. This should be included.

According to your recommendation, we performed all the statistical analysis. In univariate analysis, only the lack of cerebral metastasis was statistically associated with higher plasma concentration of nivolumab (p=0.0126). However, the results of multivariate analysis remain the same despite the inclusion of this new parameter. All the results are presented in the revised Table II. Finally, we added in the “statistical analysis” section that we tested PD-L1 expression, cerebral metastasis, tumor histology (adenocarcinoma vs squamous cell carcinoma) and stage disease (III vs IV) (lines 377-378).

 

218: 'simulation has shown'

We corrected the sentence (line 218).

Reviewer 3 Report

In this paper, Bellesoeur et al. have investigated on possible relationships between nivolumab trough levels (i.e., cmin) and clinical outcome in lung cancer patients (i.e., efficacy, toxicity).

This is a controversial and still ongoing issue because to what extent PK-guided dosing could be useful with IOD remains an ongoing story, because of conflictual reports in the literature. BMS-generated data (plus trend to use flat dosing largely exceeding the exposure levels required to ensure efficacious target engagement) suggest that PK is not an actionable item, whereas clinical reports such as Basak et al. or in silico modeling such as the Ratain group suggest that therapeutic drug monitoring could help to optimize treatments (for the Mark Ratain position, in a more pharmaco-economic perspective than to gain efficacy, still).

The presented data are much convincing and the observational study seems to have been conducted adequately. Statistical analysis looks robust. Rather surprisingly, the authors came to the opposite conclusion than Basak et al., i.e. there is no relationship between nivolumab trough levels and efficacy (the lack of rekllationship between exposure and toxicity was already shown in the Dutch study). Actually the only trend to be underlined would be a negative relationship between Cmin and survival. History of corticosteroids seems to play a major role on OS.

Because PK/PD relationships are far from bei,ng clear with IODs and with respect to teh current conflictual data tha thave been published thus far, this paper is of major interest.

The authors are required to fix the following minor issues:

The time-varying clearance they have evidenced should be better discussed in the light of the possible underlying mechanisms. in the introduction section, previous data regarding lack of PK/PD relationships with nivolumab should be discussed in the light of the metrics used - for instance, from a pharmacological point of view (or a PK/PD one), average concentrations as used in [6] barely means anything - in this respect this is not a surprise that it does not correlate with pharmacodynamic endpoints.  In addition, most data on target engagement (leading to the conclusion of flatness of PK/PD  relationships) come from evaluation of PD1 occupancy on circulating T lymphocytes, and not the infiltrated ones. Therefore, little is know about possible dose-effect relationships at the tumor level. Interpatient variability looks small on albuminemia - to what extent it does explain the little influence observed on nivolumab PK should be discussed. Albuminemia is frequently identified as a marker both for clearance and for clinical outcome (with subsequent cause VS consequence debates).

Figure 5 brings little since apparently there is no difference with or without glucocorticoids.

It is unclear why samples for cmin where withdrawn immediately after the treatment starts, and not immediately before.

Ruling out the interest of PK-dosing in the Discussion section sounds peremptory – for instance, in silico data (Goldstein et al., ASCO meeting 2019) suggest that TDM could help to postpone forthcoming courses of nivolumab, by simulating the time by which the patient is likely to be underexposed with regard to target-engagement. The presented data strongly suggest that PK-guided dosing is probably not a suitable tool to improve the efficacy of nivolumab indeed, but not that PK is totally useless with IODs.

 

 

Author Response

We are grateful for your valuable contribution to improve our manuscript.

In this paper, Bellesoeur et al. have investigated on possible relationships between nivolumab trough levels (i.e., cmin) and clinical outcome in lung cancer patients (i.e., efficacy, toxicity).

This is a controversial and still ongoing issue because to what extent PK-guided dosing could be useful with IOD remains an ongoing story, because of conflictual reports in the literature. BMS-generated data (plus trend to use flat dosing largely exceeding the exposure levels required to ensure efficacious target engagement) suggest that PK is not an actionable item, whereas clinical reports such as Basak et al. or in silico modeling such as the Ratain group suggest that therapeutic drug monitoring could help to optimize treatments (for the Mark Ratain position, in a more pharmaco-economic perspective than to gain efficacy, still).

The presented data are much convincing and the observational study seems to have been conducted adequately. Statistical analysis looks robust. Rather surprisingly, the authors came to the opposite conclusion than Basak et al., i.e. there is no relationship between nivolumab trough levels and efficacy (the lack of relationship between exposure and toxicity was already shown in the Dutch study). Actually the only trend to be underlined would be a negative relationship between Cmin and survival. History of corticosteroids seems to play a major role on OS.

Because PK/PD relationships are far from being clear with IODs and with respect to teh current conflictual data tha thave been published thus far, this paper is of major interest.

The authors are required to fix the following minor issues:

The time-varying clearance they have evidenced should be better discussed in the light of the possible underlying mechanisms.

The discussion (lines 264-269) and Figure 5 have been modified to better discuss these mechanisms.

In the introduction section, previous data regarding lack of PK/PD relationships with nivolumab should be discussed in the light of the metrics used - for instance, from a pharmacological point of view (or a PK/PD one), average concentrations as used in [6] barely means anything - in this respect this is not a surprise that it does not correlate with pharmacodynamic endpoints.  In addition, most data on target engagement (leading to the conclusion of flatness of PK/PD relationships) come from evaluation of PD1 occupancy on circulating T lymphocytes, and not the infiltrated ones. Therefore, little is know about possible dose-effect relationships at the tumor level.

We completely agree with you. A recent study suggests that a threshold value of 10 µg/mL of nivolumab would allow for >90% of maximum achievable receptor occupancy (about 80% receptor occupancy) of circulating T lymphocytes T (Ogungbenro, K et al., Clin. Pharmacol. Ther. 2018, 103, 582–590). Another study proposes a target concentration of 4.5 µg/mL (Peer CJ et al, Abstract 3115, ASCO annual meeting, Chicago 2019). In the present study, 8 patients (11.2%) and none exhibited nivolumab plasma C_min above 10 µg/mL and 4.5 µg/mL at day 14, respectively. These results could explain in part the lack of relationship between nivolumab plasma C_min and survival in our cohort. Nevertheless, one cannot exclude that this relationship is different between pharmacodynamic activity on infiltrated T lymphocytes and nivolumab concentration in the tumor microenvironment. Further investigations are required to address this issue. We now discuss this point in the revised manuscript (lines 223-231).

Interpatient variability looks small on albuminemia - to what extent it does explain the little influence observed on nivolumab PK should be discussed. Albuminemia is frequently identified as a marker both for clearance and for clinical outcome (with subsequent cause VS consequence debates).

We share your opinion. Hypoalbuminemia is a well-recognized marker of cachexia and elevated protein turnover secondary to chronic systemic inflammatory conditions, as observed in many cancer indications. In contrast with other studies (Bajaj G et al,. CPT Pharmacometrics Syst Pharmacol 2017, 6, 58–66; Hurkmans D et al., J Immunother Cancer 2019, 7, 192), we failed to identify albuminemia as an independent variability factor of nivolumab pharmacokinetics, probably because of the low interindividual variability in albuminemia. We now mention this point (lines 258-262).

Figure 5 brings little since apparently there is no difference with or without glucocorticoids.

This is because we put the wrong figure. We apologize for this mistake. The figure has been replaced with the good one. In addition, we modified the legend of Figure 5 for a better understanding.

It is unclear why samples for cmin where withdrawn immediately after the treatment starts, and not immediately before.

We apologize for the lack of clarity. Nivolumab Cmin in plasma was determined just prior to next infusion at day 14, 28 and 42. We corrected the sentence (lines 339-340).

Ruling out the interest of PK-dosing in the Discussion section sounds peremptory – for instance, in silico data (Goldstein et al., ASCO meeting 2019) suggest that TDM could help to postpone forthcoming courses of nivolumab, by simulating the time by which the patient is likely to be underexposed with regard to target-engagement. The presented data strongly suggest that PK-guided dosing is probably not a suitable tool to improve the efficacy of nivolumab indeed, but not that PK is totally useless with IODs.

We share your opinion. We modified the discussion in this direction (lines 244-250). Finally, we also corrected the conclusion (lines 391-392).